Final Report Summary - CREEPIMAGE (Development of a long term creep monitoring image based technique)
Executive Summary:
Creep is one of the most serious high-temperature damage mechanisms. It involves time-dependent deformation and high-temperature creep (HTC) cracking. It generally develops in components of engineering importance that fail over an extended time period. These include boiler superheaters and other components operating at high-temperature, petrochemical furnace and reactor vessel components and gas turbine blades. Currently, more than 70% of European power plants are more than 20-years old, 30% are more than 30 years old and 10% more than 40 years old, thus most of our power plants are approaching or have exceeded their 25-year creep design life. Safe extension of operating lifetime requires new inspection techniques that can detect in-service creep damage. Furthermore, in order to reduce emissions and improve the electricity generating efficiency, new engineering components are designed to operate at even higher temperatures and pressures. This will present severe challenges with respect to creep/fatigue life.
At present, the most commonly utilised creep measurement techniques can only be applied during outages, and consequently only about 20% of the welds can be inspected. This yields an inadequate data set for prognosis of creep life thus putting plant and Europe’s power generation infrastructure at risk.
This project aims to develop a digital image technique for long-term measurement and monitoring of creep deformation of an engineering structure/component under harsh conditions (high-temperature, irradiation, etc.), where direct sensor attachment and human access are difficult or dangerous. A digital camera equipped with a telecentric lens has been integrated into a compact prototype system with a three-legged positioning mechanism, which allows quick deployment on a power plant pipe to capture images of the component under test. Deformation pattern based digital image correlation (DPDIC) algorithms have been developed to calculate the creep deformation. A protective casing for the inspection coupon has also been designed, fabricated and trialled on an ex-service pipe. A micro laser cladding procedure has been formulated to produce a high-density grid pattern on an inspection coupon. An integrated software package has been developed, which allows the operator to acquire images, and calculate strain by DIC analysis. A remaining life prognosis method, based on the API 579 Omega model and the surface creep strain measurement has be incorporated into the system, so that high confidence in remaining life prediction of steam pipes can be achieved.
Extensive trials of the CreepImage system have been performed, which include power plant field testing to assess usability; ambient temperature testing to determine= strain measurements stability and repeatability, and high-temperature creep testing to determine strain measurement capability. The on-site usability trials resulted in the development of an enhanced image capture module (incorporating a camera positioning jig) and a documented operating procedure. Ambient temperature testing demonstrated stable and repeatable strain results (apparent strain within 100 microstrain). High-temperature testing (lasting almost two-months) demonstrated high accuracy for creep strain measurement on stainless steel 316 specimens.
Comprehensive analysis of the accelerated creep test data obtained from Inconel 625 clad P91 ex-service specimens and virgin SS316 specimens has been performed. For the SS316 accelerated creep test, good correlation was observed between the creep strain results obtained using the CreepImage system and those calculated from the elongation data recorded by the extensometers. Consequently, it was demonstrated that the CreepImage system is capable of providing accurate creep strain measurements.
Project Context and Objectives:
Power plants designed 30 to 40 years ago were intended for continuous operation, however due to their lower thermodynamic efficiency compared to newer plants they are being used for peak generation. Operating in cyclic mode (for which the plant was never designed) has serious implications for the life of components which sustain high-temperatures and high-pressures. Increasing reliance on renewables (the EC has set a binding target of 20% for electricity generation by renewables by 2020) means that cyclic operation of fossil fuel fired power plants will become the norm in the future. The reason for this is that intermittency of renewables requires sufficient standby capacity which must be cycled not only in phase with demand, but also as a function of renewable availability.
A typical electrical power-plant has around 4km of pipe-work carrying steam at pressures up to 400bar and temperatures up to 580°C. The extreme pressures produce hoop stresses in a pipe causing the pipe welds to creep continuously until creep voids coalesce to generate creep cracks which, if undiscovered, may grow until the pipe ruptures. At higher temperatures, as can occur with local overheating, deformation may be localised, with large plastic strains and local wall thinning.
The most commonly utilised creep measurement techniques can only be applied during power plant outages, and since planned outages only occur at a frequency of 1 to 4 years, only about 20% of the welds can be inspected. This yields an inadequate data set for prognosis of creep life thus putting plant and Europe’s power generation infrastructure at risk. Our solution is to develop an optical non-contact, non-intrusive, full-field strain measurement technique that can operate under these harsh conditions, and new prognosis algorithms developed to assess the remaining life of the component under test. It is based on digital image correlation. The system addresses the image distortion caused by high-temperature haze when on-site measurement is carried out on in-service steam pipes. A further challenge to be overcome is the fact that most of the components to be inspected for creep are manufactured from ferritic steels which form an oxide scale layer. A rapid test piece preparation methodology has been designed that can assess the required number of measurement points in the limited time of an outage.
The main objectives of the project are:
To develop a methodology to apply a high-temperature resistant speckle pattern to the test piece.
To develop a DPDIC algorithm to monitor creep strain via non-contact, non-intrusive measurements.
To develop algorithms to correct image distortion caused by high-temperature haze.
To develop a prognosis methodology to evaluate the remaining life of a component based on surface creep strain measurement.
To integrate the whole system and carry out field trials to validate the technique.
To develop a creep monitoring protocol for the power generation industry.
Project Results:
The main scientific and technological results arising from the project are as follows:
- In Work Package 1 preparatory research was carried out, which included sample procurement, a user requirement survey and production of the functional specifications of the CreepImage system.
Firstly, information relating to the end-user requirements was collected from all of the consortium Partners and a project advisor from the power generation industry (initially E.ON (UK) and later EDF (UK)). The current approach used for inspection of power plant pipework was reviewed, which typically follows a staged or phased approach. The initial level of inspection includes non-intrusive techniques, followed by other levels of actual monitoring, then followed by non-destructive inspections and destructive tests. This has been recognised as the most logical and cost-effective approach when performing component life assessments. Based on this information, the functional specification of the CreepImage technique was determined.
Secondly, the first iteration of the deformation pattern based digital image correlation (DPDIC) algorithm was produced. When the deformation pattern of the creep test sample is known, creep strain can be taken as the direct variable in addition to the rigid body displacement components, to represent deformation behaviour of the whole area. When a coupon is used to represent the underlying material (for the purpose of addressing oxidation of P91 material) the deformation pattern is a uniform creep strain field. With the use of a deformation pattern, the creep measurement task is turned into a pure numerical computational process, i.e. to search for a creep strain magnitude and an associated rigid body displacement field that will maximise the correlation between the original image and the inverse affine transformed image after creep deformation. It can be implemented through an optimisation procedure, and has the advantages of improved measurement accuracy and efficiency for specific measurement tasks where the deformation mode is known. The effectiveness of the DPDIC algorithm for thermal strain measurement was demonstrated by a coefficient of thermal expansion (CTE) measurement experiment on film samples.
Thirdly, a bibliographic review of four damage models was conducted. The five models were: (1) the Life Fraction model, which is time-based and often identified as Robinson’s rule; (2) the Monkman-Grant method, which makes use of the observed correlation between rupture life and creep rate; (3) the constitutive creep law models that include tertiary creep; (4) the API-MPC Omega method, which is based on tertiary creep behaviour and is often used for fitness-for service evaluations; and (5) the Continuum Damage Mechanics model, that incorporates specific damage mechanisms into a creep deformation model. For the CreepImage system, the API-MPC Omega model was selected and integrated in the prognosis module to perform remaining creep life prediction for the pipe under inspection.
- In Work Package 2 a DPDIC based strain measurement system was developed as the base-line system to operate under laboratory conditions.
Firstly, the hardware required for creep strain measurement was identified. This was based on a review of the cameras and their optic components (lenses, filters, illumination sources, etc.) for different applications.
The elementary task in digital image correlation techniques is the suitable acquisition of images. Usually, an object is illuminated by a light source and the associated images are recorded by an electronic camera. The analogue electrical image signals from the camera are converted into digital values and then stored on a computer. This step is followed by the selection and implementation of suitable image processing algorithms. Parameters that need to be determined include sensor pixel density (affects signal to noise ratio), overall pixel count, lens focal length as a function of sensor size and required angular coverage. Generally, higher fidelity images can be captured by using a larger sensor; however this requires longer focal length lenses for an equivalent field of view. Noise in the image can result in a propensity for finding false maxima in the correlation between two images, for this reason it is essential to capture images with sufficient signal to noise ratio. Illumination is also critical and can be used to overcome limitations of the sensor; however there will be limited scope for introducing complex lighting systems. The minimum f number (ratio of the focal length to the aperture) of the lens must also be considered, although there is a trade-off between aberrations and the f number. In general there will be a compromise between reducing aberrations and introducing diffraction effects. The optimum image quality will have to be determined empirically via resolution test targets, although the manufacturers data and particularly plot of sagittal and radial modulation transfer function can also be used to guide selection.
A further consideration will be whether to specify a telecentric optical system. A telecentric lens provides orthographic projection which means that the magnification is independent of distance from the lens and/or the distance from the optical axis. This means that processing the images should be simplified, however if a relatively long focal length lens is used (compared with the diagonal dimensions of the sensor) then true object space telecentricity may not be required. Based on these analyses, it was decided that the hardware for CreepImage will use a conventional camera, a Schneider Xenoplan telecentric lense and a ring light LED source as the baseline equipment.
Secondly, an optical contrast pattern was designed. This pattern must be adhered to the surface of the test-piece and deform with it to provide information for the underlying materials’ creep behavior. Digital Image Correlation (DIC) technique, being a speckle based method, enjoys the advantages of a non-contact and full-field measurement. It utilises optical images of speckle patterns on a specimen surface, and correlates sub-regions throughout the un-deformed and deformed images to obtain information for the deformation characteristics .
Generally, good speckle patterns are carriers of high information content. By applying a speckle pattern on a material, its surface becomes textured. Thus, information for pattern matching becomes available from every point of the textured surface and not just limited to a sparse grid. Furthermore, this enables the employment of a small aperture for pattern matching, known as the subset or window.
The designed speckle pattern for the CreepImage system is a geometrical array of greyvalues, in order to facilitate the acquisition of high-quality images. Specifically, the pattern is designed in a way to minimise the tendency to converge on false maxima in the correlation between the transformed images. A simulation program based on Matlab R2012a was established. It used Matlab Image Processing commands and generated random speckle patterns by a series of operations, including image expansion, image averaging/sharpening, and image shifting. The generated pattern has uniform speckle distribution resembling real laser speckle patterns. By assigning a spatial resolution with default values of 1024x1024 pixels, an 8-bit grey scale image with random geometric patterns will be generated in an uncompressed BMP format.
Following more laboratory tests, it was decided that a 2D camera system was preferable over a 3D stereo camera system. This is because in power plants, the time and working space available to carry out strain measurement is often limited. For a curved pipe surface, if the radius of curvature of the test piece exceeds a certain threshold, a correction factor will be needed to account for the curvature. Furthermore, a program to characterise the lens distortion has been implemented, and the radial distortion coefficients of the available lenses were quantified.
Thirdly, systematic comparison tests were performed in which the strain results from the DPDIC system were compared with strain gauge results. Strain measurement mean values and standard deviations of samples under rigid body motion and uniaxial extension were used to assess the sensitivity, repeatability and reliability of the DPDIC system. Results showed that the in-plane deformation of creep test specimens was measured by the CreepImage system with good accuracy. For displacement up to 2mm, a high accuracy (around 0.4µm, or 0.2pixel) can be achieved. This is an accuracy of 0.02%. For strain measurement, it is possible to obtain an accuracy of 10-20µɛ under good conditions. For strain measurement under harsh conditions (such as high-temperature), the use of a telecentric lens in conjunction with distortion correction provided a practical solution.
- In Work Package 3 the baseline CreepImage system was further developed so that it could perform strain measurements on high-temperature components.
Firstly, work was carried out to develop a methodology to apply a high-temperature resistant speckle pattern to a test piece. As most of the pipe components to be inspected for creep damage are manufactured from ferritic steels, oxidation is a major challenge. When ferritic steels operate at high-temperatures, an oxide layer is formed on the surface. Because this oxide layer is not cohesive, it is prone to spallation, making it difficult to apply an optical contrast pattern that is dimensionally stable with respect to the substrate (i.e. it only responds to deformation of the surface of the component and not to changes in surface quality or spallation). Various high-temperature coatings were tested, but the speckle pattern produced could not sustain the high-temperatures and oxidation over a prolonged time period. Hence, a methodology to attach a corrosion-resistance material to the test piece was required. The methodology developed comprised a cladding process to deposit a protective substrate on the surface of the sample and laser marking to generate a 2D pattern on the substrate.
The choice of substrate material was based on a bibliography review. The oxidation behaviour of Fe-base alloys and Ni-base alloys suggested that P91 and P92 have a good surface performance in dry environments for DIC measurement, because a protective scale of FeCr2O4 is formed at 600–700°C. This corrosion behaviour can be improved by cladding a thin foil of Fe-base alloys with a higher Cr content, such as 316SS or 304SS. In addition Ni-base alloys have appropriate corrosion behaviour, both when dry and in the presence of water vapour; although the mechanical properties of these Ni-base alloys and Fe-alloys are quite different from P91 and P92. The materials selection was made in order to allow DIC measurement, so the material with the most similar expansion coefficient and ultimate strength was required to guarantee that the substrate was dimensionally stable with respect to the parent material (i.e. it responded to deformation of the component under test and it did not generate stress during the heating cycle, or through creep difference behaviour). For P91 pipes, Inconel 625 was selected as the substrate material. For SS316 components, an red-brown oxidation layer is formed upon heating. However, the oxide layer then stabilises, hence no substrate is required for reliable DIC measurement with this material.
A micro laser cladding technique was developed to fabricate a grid/speckle pattern on a P91 test piece. Using a high-brilliance laser source it was possible to produce small features with dimensions of a few tens of micrometres. Special fluidising methods were needed to overcome the cohesive forces of the minute particles. A pulsed laser source was used to fuse the powder addition into a dot with a diameter less than 20 microns. An array of Inconel 625 dots, each having a diameter of less than 40 microns and hemispherical shape was created. The grid/speckle pattern produced can resist the effects of oxide spallation at high-temperatures.
Secondly, a technique to address image distortion caused by high-temperature haze from the test specimen was developed. Initially an image projection system was proposed to correct image distortion. However, due to the unforeseen doubling of the distortion (from pattern projection to image acquisition) and limited spatial resolution of the projector, this approach did not provide the required level of accuracy under high-temperature conditions. Consequently, an alternative approach was developed which used video image recording and an Adaptive Image Filtering (AIF) algorithm.
When a target object is subjected to high-temperature, the image intensity of every pixel will change randomly over time. However for the majority of the time, the mass points of the object will remain at fixed pixels of the image sensor. Once a sufficient number of images have been acquired as video clips (over a period of time), all distortion modes will be captured. By plotting the grey scale intensity of each single pixel taken from each independently observed frame into a frequency (probability) domain, a stochastic model can be created. Thereby, instead of assuming a normal distribution with zero mean and standard deviation, the true probability distribution of each pixel can be analysed and the statistical mode of each pixel set recorded to reconstruct the undistorted image. This is the principle of the AIF algorithm. By filtering out most of the distortion modes of a target component, an optimal image can be established as the best representation of the captured image sequence. An experimental system has been set up to implement this approach. Tests carried out have demonstrated that the AIF approach is a practical solution.
Thirdly, a high-temperature test rig based upon the Joule’s heating effect was designed and constructed. An electric current was used to heat the test specimen, however at the same time a cooling process occurred by means of heat exchange with the surrounding environment. To achieve a constant temperature it was necessary to use closed-loop feedback control to adjust the current intensity. The test rig can heat a test piece up to a temperature of 700°C and allow direct access by a digital camera to acquire the required images.
Using this rig, a series of high temperature tests were performed and video clips recorded using the prototype CreepImage system. Strain measurements obtained via post-processing of the video clips demonstrated that accurate results could be obtained (standard deviation of the difference between the horizontal and vertical strain of the specimen at 11 microstrains during the heating periods). The strain curves also showed high correlation with the temperature curves, thereby indicating reliable strain measurement results with a low uncertainty of 11 microstrains at temperatures up to 700°C.
- In Work Package 4 the hardware and software of the CreepImage system was integrated into a compact prototype system.
Firstly, an initial version of the CreepImage system was built. The opto-mechanical hardware was designed to enable rapid deployment of the system, and a thermal shielding enclosure was designed to protect the hardware from the high-temperatures that will be experienced when the system is used to monitor the creep deformation of an in-service steam pipe in a power plant.
For creep strain measurement during an outage of a power plant, key issues of the system include system positioning and illumination. A laser-based positioning and alignment module was designed to achieve accurate system positioning and optical alignment with accuracy of ±1mm. However, due to the complex nature of this design it was not implemented.
For creep strain measurement at high-temperature, thermal shielding of the CreepImage system is the major challenge. In order to minimise the time the CreepImage system is exposed to the high-temperatures in the locality of power plant steam pipes, the system has been designed to enable quick on-site set up. This is a non-trivial task as accurate creep test measurements necessitate precise positioning and alignment of the inspection equipment.
Secondly, a platform that integrates the hardware and software of the CreepImage system has been developed. This includes the Object Linking and Embedding (OLE) software and an OPC server (OLE software for Process Control) that enables interoperability of the CreepImage equipment and standardization of the control mechanism, a user-friendly human machine interface (HMI) that facilitates image acquisition, camera configuration and image visualisation, and software controls for incorporation of the image correction and DPDIC algorithms.
In order to develop the required remote control functionality in a fast and efficient manner, OPC (OLE for Process Control) software was developed. This approach is widely used by industry to facilitate communication between Microsoft Windows based programs and hardware devices.
The control software operates as a web-based application, connecting an OPC server with an OPC system service to the developed web application (the OPC client). The OPC software accepts various data inputs, translates them into an OPC format, and stores them in a common database. This process enables display and analysis of the CreepImage data. System peripherals (such as the camera module) are connected to the OPC server via the OPC tunnelling
The digital images produced are stored in the common database; the control algorithms retrieve the data from the database and perform calculations based upon the stored images. The results produced are also stored in the database and can be accessed by the user through the HMI. The CreepImage control mechanism operates the image module which comprises a camera, lens and lighting system. Management of the image module is performed remotely via the HMI which is connected to the OPC server. The HMI was designed to provide a user-friendly environment in which an authorised operator can easily connect the camera to the network, configure the camera and manage the captured images/videos. An intuitive user-interface has been designed that enables the operator to quickly become familiar with the required control and management processes. The image correction and DPDIC algorithms were integrated with the HMI using a dynamic link library (DLL). The DLL contains all of the functions necessary for pre-processing captured images (e.g. alignment of images) and to calculate the creep deformation of the specimen under test.
Thirdly, the CreepImage prototype system has been further optimised by incorporating a three-legged positioning mechanism, which enables rapid deployment of the CreepImage system and reliable image acquisition on a length of power plant steam pipe. The new version of the system was tested in the laboratory using a P91 pipe section sample, representative of the type used for high-temperature steam pipework in a power plant. Modifications to the CreepImage system hardware have enhanced both the user experience and the quality of the creep strain results obtained. Laboratory tests and demonstration of the optimised system to project Partners have shown the prototype CreepImage system to deliver the required level of creep strain measurement accuracy (>100µɛ), and to facilitate rapid deployment, in which a creep strain measurements can be performed in less than five minutes..
A protective casing has been developed that ensures the inspection coupon is protected from the harsh power plant environment, throughout the prolonged creep strain monitoring process. Two welding procedures associated with installation of the CreepImage system have been developed to provide power plant operators with the necessary assurances that the CreepImage monitoring process will not adversely affect the steam pipes under inspection.
Since P91 is prone to surface oxidation, direct application of the DIC algorithm will not produce the required level of creep strain accuracy. Consequently, a grid-based FFT method has been developed for strain measurements when the specimen surface has degraded due to prolonged exposure to high-temperatures. However, the accuracy of the strain measurements achieved is also relatively low (approximately 0.1%).
The laser micro cladding parameters has also been optimised to generate and a high-density grid pattern produced on the inspection coupon. High-temperature oven tests using specimens with this fine grid patterns showed improved creep strain measurement accuracy.
- In Work Package 5 a methodology to prognose the remaining life of a component based on surface creep strain measurements was implemented.
Firstly, a literature review of the available prognosis models was performed. The API-MPC Omega method was identified as the most suitable means of performing creep life prognosis with the CreepImage system. This model is based on the idea that the current creep strain rate, along with a brief history of creep strain rates, provides enough information to estimate the past and predict the future creep behaviour of a component. The API-MPC Omega method is included in the American Petroleum Institute Recommended Practice 579 on Fitness-for-Service document, as a reliable method of predicting creep failure in high-temperature applications using commonly used metals. One advantage of this method is that the aforementioned fitness-for-service report includes all of the material constants needed for several commonly used high-temperature application metal alloys. However, it should be noted that the report does not include values for the particular material constants needed for this application.
The concept of Omega as a material property on a par with strain rate draws attention to differences among materials in their response to creep. Some are highly strain tolerant and fail by plastic collapse after large strains, while others lose load-carrying capabilities due to creep degradation after only small strains. If a material is unstable at the microstructure level, due to precipitation, embrittlement, or accelerated softening, such behaviour can be identified from the creep curve. Hence, microstructural issues must be considered when choosing test temperatures. For materials which undergo significant dimensional changes (such as stainless steels), the advisability of applying thermal stabilising heat treatments prior to tests of virgin materials should be determined.
Advantages of the API-MPC Omega method include that it can:
• Predict the time to reach a specified strain.
• Be applied to future service life, independent of past history.
• Provide methods for dealing with tubular components rather than the uniaxial laboratory specimens.
• Allow for relatively short-time benchmark testing at temperatures much closer to operating conditions in the creep range.
• Be incorporated in finite element analysis.
However, the API-MPC Omega method also has its limitations. The integrity of the components that operate at high-temperatures and pressures is dependent on the use of design codes and assessment procedures, that ensure that the components do not fail within the required life time. These codes and procedures utilise the parent material properties of the component, in order to obtain a design or remaining safe life. Moreover, these properties are usually obtained by uniaxial testing on small scale samples of the parent material. The conditions which arise in high-temperature plant, however, differ markedly from those produced by simple uniaxial testing. In reality, components are subjected to multiaxial stresses and, because welding is often used in their fabrication, they contain weld metal and heat affected zones with properties that may differ from those of the parent material. It is therefore important that due consideration is given to the presence of weldments and multiaxial stresses when making creep life predictions
The CreepImage approach (based upon the API-MPC Omega method) has been used to predict the strain behaviour of SS316 virgin specimens and the results have been verified by experimental means. The uncertainty of the predicted rupture time is less than 2%. Hence, it was concluded that the Omega Model can predict the end-of-life of a specimen with high accuracy.
However, comparing the overall results of the API-MPC Omega method with the experimental data gave an error of 17%. A possible reason for this discrepancy is the heating and cooling cycle that was present throughout the accelerated creep test. Since the applied stress (208MPa) was well above the yield stress (205-290MPa at room temperature and around 150MPa at 620°C), repeated plastic strain occurred throughout the test (the initial test run had a plastic strain of 2.3%, and at the end of the test the total plastic strain was around 2.59%). This is likely to have had a significant detrimental effect on the life of the specimen.
Secondly, a creep monitoring protocol was established, which incorporated the CreepImage approach to inspection/creep life prognosis into the current power plant inspection strategies.
Currently, creep assessment is initiated most commonly by the client (as advised by a Technical Service Provider (TSP), or internal engineering department), as part of their asset management and predictive maintenance, or plant replacement strategies, rather than regulatory necessity. The assessment usually takes the form of a three stage process:
• Stage 1 – Desktop assessment of the power plant design and its operational parameters (usually performed at the ‘mid-life’ point of the plant).
• Stage 2 – Site measurements/confirmation of design and operational process to back-up, or confirm the Stage 1 assessment.
• Stage 3 – May involve NDT plant monitoring stress analysis and replication at a later time, unless some components have prematurely reached an advanced stage of creep damage. It should be noted that much of the Stage 3 information will be obtained during the course of statutory shutdown inspection activities, and may include replication and hardness testing.
There are no formal standards or regulations for the assessment of creep damage, or the assessment of the remaining life of a component. The purpose of the scheduled plant outages is to gather sufficient data on the condition of the high-temperature systems to satisfy all parties that the plant can be safely returned to service for a defined period of time. However, the power plant can be returned to service with restrictions on its operation in circumstances where it has not been possible to reliably underwrite the plant’s integrity for the next operating period. In such instances, a safety clause would be required, which may require that the plant is brought off-load, after a short period of time for re-inspection.
It is anticipated that the integration of the CreepImage inspection technique into existing inspection strategies will be a straightforward process. The most complex/time-consuming aspect of the CreepImage approach is the installation of the inspection coupon (that incorporates a micro gird pattern produced using the micro laser cladding process or other technique) and the installation of the protective casing used to protect the coupon from dirt/dust or damage during the period of inspection (typically inspection will be performed over a number of years). The installation of a single coupon and its associated protective case (via suitably qualified welding procedures) is expected to take around one-hour, depending on the accessibility of the inspection area. An initial image/video clip will then be recorded using the specifically designed image acquisition jig. This process will take less than five minutes (including laptop/hardware setup). Once the installation and initial inspection process has been performed (during a planned outage), subsequent inspections can be carried out either during power plant operation or during subsequent outages. Similarly, this simple image/video capture process will be performed in less than five minutes, once access to the inspection area has been achieved. Integration of the CreepImage inspection process into existing inspection strategies will be further simplified by performing the strain calculations for a number of inspection areas, via off-line batch data processing away from the power plant, following the data acquisition procedure.
The types of componet that can be monitored using the CreepImage system has also been defined. The most informative assessment of component condition is obtained during an outage. Components typically assessed for creep damage (operating at temperatures >400°C) include:
• Main steam and hot reheat pipework systems.
• Header systems, manifolds and drains.
• Valves and steam chests.
• A wide range of welds (pipe-to-pipe, large branch, pipe/header penetrations, attachments).
• High pressure and intermediate pressure turbines (rotors, casings, etc.).
• Any component identified as having high stress levels via finite element analysis or other theoretical calculation.
The CreepImage system is capable of inspecting all of the above components provided there is sufficient surface area to attach the inspection coupon, and there is adequate access to the component (the CreepImage optical system requires a 1m stand-off distance from the component under test).
An operating procedure for the CreepImage system that is compatible with current power plant inspection strategies has been formulated. The CreepImage implementation strategy is focused on inspection services during a power plant outage. However, in-service inspection is not excluded, provided that all safety regulations regarding high-temperature working are met. The CreepImage operating procedure is defined as follows:
1. Obtain permission to perform on-site inspection from the relevant authority.
2. Ensure appropriate personal protective equipment (PPE) is worn by all inspection personnel.
3. Erect barriers and signs to prevent unauthorised personnel entering the inspection area.
4. Perform a risk assessment of the working environment and take necessary actions to minimise risk.
5. Initial inspection (performed during a power plant outage):
a) Identify appropriate inspection sites according to the requirements of the power plant operator (in conjunction with theoretical/simulation analyses). Inspection sites are usually situated around welds and incorporate the heat affected zone.
b) Ensure there is adequate access to the component under test.
c) Expose the surface of the component to be inspected (e.g. remove insulation from pipework).
d) Weld an Inconel 625 coupon incorporating the speckle/grid pattern to the area of the component to be inspected.
e) Weld the protective casing to the surface of the component to be inspected (this also ensures that the camera is correctly positioned throughout the entire inspection process).
f) If necessary, apply post weld heat treatment (PWHT) to release weld residual stress.
6. Coupon inspection (performed during an outage, or whilst the power plant is operational):
a) Ensure there is adequate access to the component under test.
b) Expose the surface of the component to be inspected (e.g. remove insulation from pipework). Particular care must be taken with hot components during in-service power plant inspection.
c) Check if the speckle/grid pattern has remained intact.
7. Image acquisition:
a) Connect the camera to the laptop.
b) Mount the high-definition camera (with integral telecentric lens and LED lighting) in the positioning holes in the top of the protective casing.
c) Run the CreepImage image capture software and observe a clear in-focus image of the area under test on the laptop screen.
d) Select the Capture Image option to capture a still image or a video clip of the component under test.
e) Replace any component covering removed prior to inspection (e.g. replace insulation on pipework).
f) Use the CreepImage software to enhance the clarity of the captured speckle pattern and aid image correlation (if required).
8. Subsequent inspections:
a) Run the DIC software (either on-site or after leaving the site) to obtain the accumulated creep strain value.
b) Enter the accumulated creep strain result into the creep life prognosis model to perform a remaining life assessment for the component under test.
c) Export the CreepImage Report using the reporting tool integrated in the CreepImage software.
After an appropriate time interval (6-18 months) repeat steps 1-8 to capture additional speckle pattern images.
The information included in the CreepImage Inspection Report can be divided into three categories:
1. General information on the inspection (input by the user).
2. The CreepImage analysis results (produced by the software).
3. The operator’s comments (input by the service provider).
More specifically, the information included in the report includes:
• Inspection date.
• CreepImage version number (to record any post-project changes to the system specification (e.g. a different hardware setup, DIC algorithm, or methodology).
• Name of the power plant under inspection.
• Name of the service provider and the operator’s name.
• Information relating to the component under inspection:
o Component ID
o Unit
o Weld
o Weld filler
o Component material
o Component/weld service duty
• Results of the CreepImage analysis:
o DPDIC
o Horizontal/vertical/shearing strain
o Strain rate
o Omega value for the creep life prognosis
o In-service/operational time of the component
o Creep life prognosis results
• A ‘free text’ section, where the operator can enter any additional comments (e.g. possible recommendations, or resulting actions).
• One or two digital images of the component under inspection (e.g. photographs or engineering drawings) for identification purposes.
Health and safety issues are of pramount importance in a power plant. Consequently, a Method Statement and Risk Assessment for the prototype CreepImage system have been produced (these two documents typically make up the workplace safety plan). In combination, they provide detailed guidance on how to carry out a CreepImage inspection safely. These documents were specifically developed in order to gain permission from the power plant operator to perform the CreepImage field trials. They must be referenced during a safety induction and then as required throughout the inspection process. The CreepImage Method Statement gives specific instructions on how to safely perform a creep strain measurement using the CreepImage prototype system in an operational power plant. The Risk Assessment determines quantitatively the risks related to the hazards associated with carrying out a CreepImage inspection, as well as the likelihood of them occurring. It ensures that all risks are understood and if they are acceptable.
- In Work Package 6 comprehensive system testing was performed to demonstrate that component life can be predicted with sufficient confidence to minimise the likelihood of catastrophic failure.
Firstly, extensive trials of the CreepImage system were performed including: power plant field testing to assess usability; ambient temperature testing to determined strain measurement stability and repeatability, and a high-temperature creep testing to determine strain measurement capability.
During the on-site preliminary trials of the CreepImage system (performed in a power station during a scheduled outage) two pipe components were examined. The first was a main valve with a complex geometry, which was used to control the flow of a high-temperature/high pressure steam pipe. The major problem encountered inspecting this component was the limited access and the instability of the surrounding floor. In fact, due to the severely restricted access, it was concluded that it would be impossible to deploy the tripod/camera system in a location suitable for reliable image acquisition. The second location chosen for inspection was a temporary wooden platform positioned on top of a scaffold tower. The scaffolding had been erected to facilitate inspection of a length of main steam pipe, which had already had its lagging removed. The CreepImage system (second version prototype, mounted on a tripod) was setup. No mains power was necessary for this trial, as the camera was powered by a USB cable connected to the laptop, the laptop used battery power, and the ambient light was considered sufficient for the camera, hence the LED ring-light was not activated. Prior to the start of the trial inspection, a speckle pattern was spray painted on to a small area of the pipe (this test was designed to have minimal impact on the power plant infrastructure). Then following initiation of the CreepImage image capture software, live images of the speckle pattern were displayed on the laptop screen. The relative motion between the pipe and the camera was clearly seen. To determine the magnitude of the vibration, two video sequences were recorded (with a frame rate of 11fps). Each video clip lasted approximately 30 seconds. Post-processing of the video clips (performed off-site) showed that the typical amplitude of the vibration was approximately 0.2mm. However, the vibration amplitude reached approximately 0.5mm when nearby engineering works involved a sudden impact. Vibrations of this amplitude were shown not to affect the performance of the CreepImage system. The usability of the CreepImage system was evaluated via these tests.
It should be noted that a full CreepImage field trial would have involved the attachment of an inspection coupon and a protective casing (to prevent contamination of the coupon) to the surface of the component under test. However, due to the strict requirement to ensure that inspection process did not interfere with the power plant infrastructure, this was not possible. Laser welding is recognised as the most appropriate technique for welding thin sheet to thick section materials. However, laser welding is not a portable process and therefore it is unsuitable for use in an operational power plant. Consequently, the portable and compact TIG welding process was identified as the joining technique best suited for this application. In order to gain permission for a test that involves welding onto a steam pipe it will be necessary to demonstrate to the power plant operator that the welding operations do not affect the structural integrity of the steam pipe. Hence, qualified welding procedures must be developed to ensure that any residual stresses do not adversely affect the pipework. Two TIG welding procedures (for the attachment of the inspection coupon and the protective casing to a P91 pipe) were developed as part of the project. However, further work is required to fully qualify the procedures, in order that they may be accepted by a power plant operator.
Ambient temperature laboratory trials were performed to demonstrate the stability and repeatability of the CreepImage system. The trials started with the installation of an Inconel 625 inspection coupon and a protective casing on an ex-service P91 pipe specimen. The image capture module was then located in the positioning holes in the top of the protective casing. This ensured that the correct camera orientation and stand-off distance. Once the image capture module was correctly positioned, the image acquisition process was initiated and a series of video clips (lasting approximately 30 seconds) were captured. The video clips were then post-processed to calculate the apparent strain of the coupon. Results showed that although the fluctuation in apparent strain was around ±150microstrain, the mean strain was less than 20microstrain. Hence, high strain measurement accuracy can be achieved by simply averaging multiple measurement results.
High-temperature laboratory trials were also conducted, which validated the strain measurement capability of the CreepImage system on SS316 over a prolonged period. The procedure for the oven creep test was as follows:
1) Fabricate the creep test specimen and produce the required grid pattern on its surface using the micro laser cladding process. The grid pattern consists of a series of evenly distributed small dots of Inconel 625 (dot separation 0.1mm dot diameter 0.1mm).
2) Capture an initial image of the grid pattern using the CreepImage system.
3) Weld three thermocouples on to the side of the specimen (top, middle and bottom positioning). The three thermocouples are used to monitor the temperature distribution across the specimen.
4) Install the specimen and load frame in the creep test oven and apply the loading.
5) Connect the two extensometers to the control computer.
6) Start the test by heating the specimen to the predefined temperature (e.g. 620°C).
7) Once the required temperature is reached, apply the required load (calculated to produce the required tensile stress in the specimen).
8) After a predefined time, stop the test by removing the load and switching-off the oven.
9) Once the oven has cooled down to ambient temperature, move the oven to expose the specimen to the CreepImage camera.
10) Capture additional images of the grid pattern using the CreepImage system. Note, due to access restrictions around the oven the specimen and the loading frame were removed from the oven for the image acquisition process.
11) Go to Step 4 to resume the test (repeat this sequence until the specimen ruptures).
Ideally, the high-temperature creep test would have been performed using an oven with a glass viewing window. With such an oven, digital images could have been acquired throughout the test without interfering with the specimen. However, as this type of oven was not available, a conventional creep test oven was used. The main drawback to this approach was that it was necessary to stop the test and wait for the specimen to cool down before image acquisition could take place. However, an advantage of this approach is that the heating and cooling cycle resembles the temperature variation experienced by a coal fired power plant’s steam pipes, as the plant is brought on and off-line to match varying output demand (there may be around 250 temperature fluctuations in one-year for a typical coal fired power generator). It should be noted that this temperature fluctuation is one of the main reasons why high-temperature paint is unsuitable for producing the required DIC grid/speckle pattern on a power plant’s steam pipework over the long-term.
An accelerated creep test was performed on a SS316 specimen with a grid pattern superimposed on its surface. During the test, the specimen was subjected a tensile stress of 208MPa at 620°C. Three thermocouples and two extensometers were attached to the specimen to monitor the temperature and elongation. The elongation data was used to calculate the creep strain between different test runs, and the results were compared with the strain measurements made by the CreepImage system. A total of 15 test runs (with durations ranging from seven to 94 hours) were performed. Although the entire test had a duration of just 528 hours (at full load and temperature), the actual time to perform the trial was almost two months, due to the need to temporarily halt the creep test (to allow the specimen to cool down to ambient temperature) each time a DIC measurement was performed.
It should be noted that the DIC technique can only measure the total remaining strain of a creep test specimen, and it is not possible to differentiate between creep strain and plastic strain from the total remaining strain. However, it is possible to use the elongation data recorded by the test machine to detect which part of the total strain is associated with creep deformation. This is because creep strain is a time dependent deformation that typically occurs at high-temperatures.
After processing all of the captured images and elongation data, the creep strain and strain rate of the specimen were obtained. The strain rate was high in the primary creep stage, it then gradually decreased, and stabilised at approximately 1.8x10-4/h from 40 to 250h. It then increased from 250h until its final rupture at around 530h.
Since a considerable number of small cracks were found on the specimen surface at 528h, the test was terminated without actually breaking the specimen. This decision was made because it was believed that the specimen was only a few hours away from final rupture (the accumulated creep strain was calculated at 17.71% and the strain rate increased from the steady state rate of 1.8x10-4/h to 6.34x10-4/h). The gauge length of the specimen after Test Run 15 was measured to have an extension of 7.1mm. Since the original gauge length was 35mm, this extension gave an overall tensile strain of 20.3%. The difference of 2.59% between total strain and total creep strain corresponded to the plastic deformation. However, the accumulated total remaining strain from the DIC measurement was 20.11%. This value is very close to the total tensile strain of 20.3% (determined by the gauge length measurements), which means an error of only 1%. In view of the fact that the material beyond the gauge length of the specimen also deformed during the creep test, the overall tensile strain value of 20.3% is likely to be an overestimate. Therefore, it was concluded that the accumulated remaining strain value of 20.11% (obtained from the DIC technique) most accurately represents the actual total deformation of the specimen. This demonstrates the high accuracy for creep strain measurement of the CreepImage system.
Secondly, comprehensive data analysis was performed on the strain measurements obtained using the CreepImage system on a virgin SS316 specimen during an accelerated creep test at 620°C. High correlation (with an R-squared value of 0.9996) was observed between the strain results obtained using the CreepImage system and the creep strain calculated from the elongation data recorded using extensometers. Consequently, it was concluded that the CreepImage system is capable of providing highly accurate creep strain measurements on SS316.
Although it was not possible to perform a complete creep strain measurement in a power plant within the two-year duration of the project, preparations have been made to start a full CreepImage creep strain field trial, during the next scheduled outage at the EDF West Burton Power Station.
Potential Impact:
The final results are a creep strain measurement system and a remaining life prognosis methodology. These include (1) DPDIC algorithm for creep strain measurement; (2) Image correction algorithm to correct distortion caused by high temperature haze; (3) Integrated CreepImage system for creep measurement during power plant outages and whilst a power plant is in operation; (4) Remaining life prognosis methodology based on the API 579 Omega model using creep strain measurements; and (5) Trial test results.
Potential impacts from the project include increased probability of identification and detection of creep damage in power generation (both thermal and nuclear) and petrochemical processing plants. This will decrease the probability of failure of high temperature components due to creep damage and type IV cracking in particular. These types of failures not only result in costly enforced outages, they are frequently accompanied by potential risk to human safety. The optical strain measurement system has wide applicability. In particular gas turbines, steam turbines and steam components could benefit. Hence, the structural integrity of such systems and components will be maintained in order to reduce the likelihood of catastrophic failure.
A series of appropriate dissemination activities were identified and performed by the project Partners. A project website comprising a public area for the dissemination of information, and password protected members area for confidential project files has been created (www.creepimage.eu ). An article that details the aims, approach and anticipated impacts of the CreepImage project has appeared in TWI’s Connect magazine (Issue 185 - July/August 2013). A CreepImage article has also appeared on the website of the UK’s Energy Generation and Supply Knowledge Transfer Network. A conference paper was produced by the project Partners TWI, Dantec Dynamics and Aimen for the British Society for Strain Measurement's (BSSM) 9th International Conference on Advances in Experimental Mechanics, held at Cardiff University, 3rd-5th September 2013. A three-minute promotional video has been produced for public release. It can be viewed on the CreepImage website (www.creepimage.eu/publication) or on the YouTube website at www.youtube.com/watch?v=8jlutNEF2Yg.
To date three major dissemination activities have been performed to promote the CreepImage project. A fourth will be carried out on the 31st October 2013. A presentation was delivered at the EDF West Burton Power Plant in order to gain permission to perform the essential on-site field trials; a conference paper was presented at the 9th International Conference on Advances in Experimental Mechanics; the CreepImage system was promoted at the BINDT NDT 2013 Conference; and the CreepImage system will be demonstrated at the R4i Research for Impact Event.
Patent searches performed during the project term resulted in no results relating to high temperature creep strain measurement via DIC. However, Chinese Patent CN 201795989U registered the idea of attaching a coupon to a base material for the purpose of creep strain measurement. It covers the concept of a protective casing and digital image analysis, but not the use of a grid pattern, nor the use of a telecentric lens for image acquisition to minimise the apparent strain which is inevitable if a conventional lens were used to inspect a curved surface on a high temperature component. It is considered that this patented idea cannot achieve accurate creep strain measurements claimed by the patent owners. Moreover, extensive literature searches have found no evidence that the patented ideas have been practically tested.
The patent searches and literature surveys performed during the project have demonstrated that the CreepImage SME Partners could potentially submit a number of patents relating to the advancements made in the CreepImage project.
Possible patent applications could cover:
• High temperature strain measurement algorithms.
• Distortion correction algorithms.
• Pattern production methods for optical measurements.
• The entire CreepImage system (assembled components).
However, at the present time there are no plans for the SME Partners to apply for a patent, as Partner discussions have identified the need to further investigate possible markets and to arrange more extensive field trials to fully validate the inspection technique. It should be noted that a full creep strain monitoring trial (performed in an operational power plant) would have a minimum duration of several years, due to the prolonged nature of creep damage formation and the lengthy interval between successive power plant outages (typically two years for a minor outage and four years for a major outage). Nevertheless, the SME Partners will still be in a position to benefit from the CreepImage developments through their increased know-how of image analysis, DIC techniques and remaining life prognosis methodologies.
Overall Strategy for Exploitation
The exploitation and commercialisation of the CreepImage system will occur in several distinct phases, covering product development, initial exploitation, European expansion and global expansion. The exploitation strategy following project closure is outlined below.
Phase 1: Demonstration of the capability of the CreepImage system to predict the remaining life of pipework during power plant outages (years 1-4)
1. The first demonstration of the CreepImage system performing creep strain measurements to potential customers will be initiated by installing the inspection coupon and associated protective casing on a steam pipe during a scheduled outage of a power plant. A subsequent scheduled outage is then required to obtain a second image of the coupon at a similar ambient temperature. In this way, the creep strain accumulated between the two outages will be measured and the creep strain rate calculated. In addition to the measurements made during outages, measurements in an operational power plant will be performed following the first outage. However this process will be significantly more complex, due to the additional risks involved with working close to high-temperature steam pipes.
2. Comparison between conventional creep strain and CreepImage methods.
3. Improvements to the CreepImage prototype.
4. Improvements to the CreepImage operating procedures.
Phase 2: Development of an Industrial System (years 1-2)
1. Ensure the prototype is physically robust and suitable for the harsh environment of a power plant.
2. Development of a user-friendly and easy to use software interface (suitable for a technician to operate).
3. Refinement of the operating procedures, so that they can be easily adopted/integrated into the power plant operator’s working practices.
Phase 3: Demonstration of the capability of the CreepImage system to predict the remaining life of pipework in an operational power plant (years 2-7)
1. Development of methods to perform creep strain measurements on high-temperature components.
2. Development of a heat protection system for the CreepImage equipment.
3. Development of operating procedures for high-temperature environments.
4. Further develop the CreepImage creep life prognosis model for high-temperature components.
Phase 4: Industrialisation of a High Temperature System (years 6-7)
1. Making the Phase 2 system sufficiently robust for the high temperature environment of a power plant.
2. The implementation of any necessary software updates.
3. The creation of operating procedures that can be easily adopted by the power plant operators/inspection providers.
List of Websites:
www.creepimage.eu
Contact details of the project's coordinator:
Jianxin Gao
TWI Limited,
Granta Park, Great Abington
Cambridge, CB21 6AL
UK
Tel: +44 1223 899000
Fax: +44 1223 892588
E-mail: jianxin.gao@twi.co.uk
Creep is one of the most serious high-temperature damage mechanisms. It involves time-dependent deformation and high-temperature creep (HTC) cracking. It generally develops in components of engineering importance that fail over an extended time period. These include boiler superheaters and other components operating at high-temperature, petrochemical furnace and reactor vessel components and gas turbine blades. Currently, more than 70% of European power plants are more than 20-years old, 30% are more than 30 years old and 10% more than 40 years old, thus most of our power plants are approaching or have exceeded their 25-year creep design life. Safe extension of operating lifetime requires new inspection techniques that can detect in-service creep damage. Furthermore, in order to reduce emissions and improve the electricity generating efficiency, new engineering components are designed to operate at even higher temperatures and pressures. This will present severe challenges with respect to creep/fatigue life.
At present, the most commonly utilised creep measurement techniques can only be applied during outages, and consequently only about 20% of the welds can be inspected. This yields an inadequate data set for prognosis of creep life thus putting plant and Europe’s power generation infrastructure at risk.
This project aims to develop a digital image technique for long-term measurement and monitoring of creep deformation of an engineering structure/component under harsh conditions (high-temperature, irradiation, etc.), where direct sensor attachment and human access are difficult or dangerous. A digital camera equipped with a telecentric lens has been integrated into a compact prototype system with a three-legged positioning mechanism, which allows quick deployment on a power plant pipe to capture images of the component under test. Deformation pattern based digital image correlation (DPDIC) algorithms have been developed to calculate the creep deformation. A protective casing for the inspection coupon has also been designed, fabricated and trialled on an ex-service pipe. A micro laser cladding procedure has been formulated to produce a high-density grid pattern on an inspection coupon. An integrated software package has been developed, which allows the operator to acquire images, and calculate strain by DIC analysis. A remaining life prognosis method, based on the API 579 Omega model and the surface creep strain measurement has be incorporated into the system, so that high confidence in remaining life prediction of steam pipes can be achieved.
Extensive trials of the CreepImage system have been performed, which include power plant field testing to assess usability; ambient temperature testing to determine= strain measurements stability and repeatability, and high-temperature creep testing to determine strain measurement capability. The on-site usability trials resulted in the development of an enhanced image capture module (incorporating a camera positioning jig) and a documented operating procedure. Ambient temperature testing demonstrated stable and repeatable strain results (apparent strain within 100 microstrain). High-temperature testing (lasting almost two-months) demonstrated high accuracy for creep strain measurement on stainless steel 316 specimens.
Comprehensive analysis of the accelerated creep test data obtained from Inconel 625 clad P91 ex-service specimens and virgin SS316 specimens has been performed. For the SS316 accelerated creep test, good correlation was observed between the creep strain results obtained using the CreepImage system and those calculated from the elongation data recorded by the extensometers. Consequently, it was demonstrated that the CreepImage system is capable of providing accurate creep strain measurements.
Project Context and Objectives:
Power plants designed 30 to 40 years ago were intended for continuous operation, however due to their lower thermodynamic efficiency compared to newer plants they are being used for peak generation. Operating in cyclic mode (for which the plant was never designed) has serious implications for the life of components which sustain high-temperatures and high-pressures. Increasing reliance on renewables (the EC has set a binding target of 20% for electricity generation by renewables by 2020) means that cyclic operation of fossil fuel fired power plants will become the norm in the future. The reason for this is that intermittency of renewables requires sufficient standby capacity which must be cycled not only in phase with demand, but also as a function of renewable availability.
A typical electrical power-plant has around 4km of pipe-work carrying steam at pressures up to 400bar and temperatures up to 580°C. The extreme pressures produce hoop stresses in a pipe causing the pipe welds to creep continuously until creep voids coalesce to generate creep cracks which, if undiscovered, may grow until the pipe ruptures. At higher temperatures, as can occur with local overheating, deformation may be localised, with large plastic strains and local wall thinning.
The most commonly utilised creep measurement techniques can only be applied during power plant outages, and since planned outages only occur at a frequency of 1 to 4 years, only about 20% of the welds can be inspected. This yields an inadequate data set for prognosis of creep life thus putting plant and Europe’s power generation infrastructure at risk. Our solution is to develop an optical non-contact, non-intrusive, full-field strain measurement technique that can operate under these harsh conditions, and new prognosis algorithms developed to assess the remaining life of the component under test. It is based on digital image correlation. The system addresses the image distortion caused by high-temperature haze when on-site measurement is carried out on in-service steam pipes. A further challenge to be overcome is the fact that most of the components to be inspected for creep are manufactured from ferritic steels which form an oxide scale layer. A rapid test piece preparation methodology has been designed that can assess the required number of measurement points in the limited time of an outage.
The main objectives of the project are:
To develop a methodology to apply a high-temperature resistant speckle pattern to the test piece.
To develop a DPDIC algorithm to monitor creep strain via non-contact, non-intrusive measurements.
To develop algorithms to correct image distortion caused by high-temperature haze.
To develop a prognosis methodology to evaluate the remaining life of a component based on surface creep strain measurement.
To integrate the whole system and carry out field trials to validate the technique.
To develop a creep monitoring protocol for the power generation industry.
Project Results:
The main scientific and technological results arising from the project are as follows:
- In Work Package 1 preparatory research was carried out, which included sample procurement, a user requirement survey and production of the functional specifications of the CreepImage system.
Firstly, information relating to the end-user requirements was collected from all of the consortium Partners and a project advisor from the power generation industry (initially E.ON (UK) and later EDF (UK)). The current approach used for inspection of power plant pipework was reviewed, which typically follows a staged or phased approach. The initial level of inspection includes non-intrusive techniques, followed by other levels of actual monitoring, then followed by non-destructive inspections and destructive tests. This has been recognised as the most logical and cost-effective approach when performing component life assessments. Based on this information, the functional specification of the CreepImage technique was determined.
Secondly, the first iteration of the deformation pattern based digital image correlation (DPDIC) algorithm was produced. When the deformation pattern of the creep test sample is known, creep strain can be taken as the direct variable in addition to the rigid body displacement components, to represent deformation behaviour of the whole area. When a coupon is used to represent the underlying material (for the purpose of addressing oxidation of P91 material) the deformation pattern is a uniform creep strain field. With the use of a deformation pattern, the creep measurement task is turned into a pure numerical computational process, i.e. to search for a creep strain magnitude and an associated rigid body displacement field that will maximise the correlation between the original image and the inverse affine transformed image after creep deformation. It can be implemented through an optimisation procedure, and has the advantages of improved measurement accuracy and efficiency for specific measurement tasks where the deformation mode is known. The effectiveness of the DPDIC algorithm for thermal strain measurement was demonstrated by a coefficient of thermal expansion (CTE) measurement experiment on film samples.
Thirdly, a bibliographic review of four damage models was conducted. The five models were: (1) the Life Fraction model, which is time-based and often identified as Robinson’s rule; (2) the Monkman-Grant method, which makes use of the observed correlation between rupture life and creep rate; (3) the constitutive creep law models that include tertiary creep; (4) the API-MPC Omega method, which is based on tertiary creep behaviour and is often used for fitness-for service evaluations; and (5) the Continuum Damage Mechanics model, that incorporates specific damage mechanisms into a creep deformation model. For the CreepImage system, the API-MPC Omega model was selected and integrated in the prognosis module to perform remaining creep life prediction for the pipe under inspection.
- In Work Package 2 a DPDIC based strain measurement system was developed as the base-line system to operate under laboratory conditions.
Firstly, the hardware required for creep strain measurement was identified. This was based on a review of the cameras and their optic components (lenses, filters, illumination sources, etc.) for different applications.
The elementary task in digital image correlation techniques is the suitable acquisition of images. Usually, an object is illuminated by a light source and the associated images are recorded by an electronic camera. The analogue electrical image signals from the camera are converted into digital values and then stored on a computer. This step is followed by the selection and implementation of suitable image processing algorithms. Parameters that need to be determined include sensor pixel density (affects signal to noise ratio), overall pixel count, lens focal length as a function of sensor size and required angular coverage. Generally, higher fidelity images can be captured by using a larger sensor; however this requires longer focal length lenses for an equivalent field of view. Noise in the image can result in a propensity for finding false maxima in the correlation between two images, for this reason it is essential to capture images with sufficient signal to noise ratio. Illumination is also critical and can be used to overcome limitations of the sensor; however there will be limited scope for introducing complex lighting systems. The minimum f number (ratio of the focal length to the aperture) of the lens must also be considered, although there is a trade-off between aberrations and the f number. In general there will be a compromise between reducing aberrations and introducing diffraction effects. The optimum image quality will have to be determined empirically via resolution test targets, although the manufacturers data and particularly plot of sagittal and radial modulation transfer function can also be used to guide selection.
A further consideration will be whether to specify a telecentric optical system. A telecentric lens provides orthographic projection which means that the magnification is independent of distance from the lens and/or the distance from the optical axis. This means that processing the images should be simplified, however if a relatively long focal length lens is used (compared with the diagonal dimensions of the sensor) then true object space telecentricity may not be required. Based on these analyses, it was decided that the hardware for CreepImage will use a conventional camera, a Schneider Xenoplan telecentric lense and a ring light LED source as the baseline equipment.
Secondly, an optical contrast pattern was designed. This pattern must be adhered to the surface of the test-piece and deform with it to provide information for the underlying materials’ creep behavior. Digital Image Correlation (DIC) technique, being a speckle based method, enjoys the advantages of a non-contact and full-field measurement. It utilises optical images of speckle patterns on a specimen surface, and correlates sub-regions throughout the un-deformed and deformed images to obtain information for the deformation characteristics .
Generally, good speckle patterns are carriers of high information content. By applying a speckle pattern on a material, its surface becomes textured. Thus, information for pattern matching becomes available from every point of the textured surface and not just limited to a sparse grid. Furthermore, this enables the employment of a small aperture for pattern matching, known as the subset or window.
The designed speckle pattern for the CreepImage system is a geometrical array of greyvalues, in order to facilitate the acquisition of high-quality images. Specifically, the pattern is designed in a way to minimise the tendency to converge on false maxima in the correlation between the transformed images. A simulation program based on Matlab R2012a was established. It used Matlab Image Processing commands and generated random speckle patterns by a series of operations, including image expansion, image averaging/sharpening, and image shifting. The generated pattern has uniform speckle distribution resembling real laser speckle patterns. By assigning a spatial resolution with default values of 1024x1024 pixels, an 8-bit grey scale image with random geometric patterns will be generated in an uncompressed BMP format.
Following more laboratory tests, it was decided that a 2D camera system was preferable over a 3D stereo camera system. This is because in power plants, the time and working space available to carry out strain measurement is often limited. For a curved pipe surface, if the radius of curvature of the test piece exceeds a certain threshold, a correction factor will be needed to account for the curvature. Furthermore, a program to characterise the lens distortion has been implemented, and the radial distortion coefficients of the available lenses were quantified.
Thirdly, systematic comparison tests were performed in which the strain results from the DPDIC system were compared with strain gauge results. Strain measurement mean values and standard deviations of samples under rigid body motion and uniaxial extension were used to assess the sensitivity, repeatability and reliability of the DPDIC system. Results showed that the in-plane deformation of creep test specimens was measured by the CreepImage system with good accuracy. For displacement up to 2mm, a high accuracy (around 0.4µm, or 0.2pixel) can be achieved. This is an accuracy of 0.02%. For strain measurement, it is possible to obtain an accuracy of 10-20µɛ under good conditions. For strain measurement under harsh conditions (such as high-temperature), the use of a telecentric lens in conjunction with distortion correction provided a practical solution.
- In Work Package 3 the baseline CreepImage system was further developed so that it could perform strain measurements on high-temperature components.
Firstly, work was carried out to develop a methodology to apply a high-temperature resistant speckle pattern to a test piece. As most of the pipe components to be inspected for creep damage are manufactured from ferritic steels, oxidation is a major challenge. When ferritic steels operate at high-temperatures, an oxide layer is formed on the surface. Because this oxide layer is not cohesive, it is prone to spallation, making it difficult to apply an optical contrast pattern that is dimensionally stable with respect to the substrate (i.e. it only responds to deformation of the surface of the component and not to changes in surface quality or spallation). Various high-temperature coatings were tested, but the speckle pattern produced could not sustain the high-temperatures and oxidation over a prolonged time period. Hence, a methodology to attach a corrosion-resistance material to the test piece was required. The methodology developed comprised a cladding process to deposit a protective substrate on the surface of the sample and laser marking to generate a 2D pattern on the substrate.
The choice of substrate material was based on a bibliography review. The oxidation behaviour of Fe-base alloys and Ni-base alloys suggested that P91 and P92 have a good surface performance in dry environments for DIC measurement, because a protective scale of FeCr2O4 is formed at 600–700°C. This corrosion behaviour can be improved by cladding a thin foil of Fe-base alloys with a higher Cr content, such as 316SS or 304SS. In addition Ni-base alloys have appropriate corrosion behaviour, both when dry and in the presence of water vapour; although the mechanical properties of these Ni-base alloys and Fe-alloys are quite different from P91 and P92. The materials selection was made in order to allow DIC measurement, so the material with the most similar expansion coefficient and ultimate strength was required to guarantee that the substrate was dimensionally stable with respect to the parent material (i.e. it responded to deformation of the component under test and it did not generate stress during the heating cycle, or through creep difference behaviour). For P91 pipes, Inconel 625 was selected as the substrate material. For SS316 components, an red-brown oxidation layer is formed upon heating. However, the oxide layer then stabilises, hence no substrate is required for reliable DIC measurement with this material.
A micro laser cladding technique was developed to fabricate a grid/speckle pattern on a P91 test piece. Using a high-brilliance laser source it was possible to produce small features with dimensions of a few tens of micrometres. Special fluidising methods were needed to overcome the cohesive forces of the minute particles. A pulsed laser source was used to fuse the powder addition into a dot with a diameter less than 20 microns. An array of Inconel 625 dots, each having a diameter of less than 40 microns and hemispherical shape was created. The grid/speckle pattern produced can resist the effects of oxide spallation at high-temperatures.
Secondly, a technique to address image distortion caused by high-temperature haze from the test specimen was developed. Initially an image projection system was proposed to correct image distortion. However, due to the unforeseen doubling of the distortion (from pattern projection to image acquisition) and limited spatial resolution of the projector, this approach did not provide the required level of accuracy under high-temperature conditions. Consequently, an alternative approach was developed which used video image recording and an Adaptive Image Filtering (AIF) algorithm.
When a target object is subjected to high-temperature, the image intensity of every pixel will change randomly over time. However for the majority of the time, the mass points of the object will remain at fixed pixels of the image sensor. Once a sufficient number of images have been acquired as video clips (over a period of time), all distortion modes will be captured. By plotting the grey scale intensity of each single pixel taken from each independently observed frame into a frequency (probability) domain, a stochastic model can be created. Thereby, instead of assuming a normal distribution with zero mean and standard deviation, the true probability distribution of each pixel can be analysed and the statistical mode of each pixel set recorded to reconstruct the undistorted image. This is the principle of the AIF algorithm. By filtering out most of the distortion modes of a target component, an optimal image can be established as the best representation of the captured image sequence. An experimental system has been set up to implement this approach. Tests carried out have demonstrated that the AIF approach is a practical solution.
Thirdly, a high-temperature test rig based upon the Joule’s heating effect was designed and constructed. An electric current was used to heat the test specimen, however at the same time a cooling process occurred by means of heat exchange with the surrounding environment. To achieve a constant temperature it was necessary to use closed-loop feedback control to adjust the current intensity. The test rig can heat a test piece up to a temperature of 700°C and allow direct access by a digital camera to acquire the required images.
Using this rig, a series of high temperature tests were performed and video clips recorded using the prototype CreepImage system. Strain measurements obtained via post-processing of the video clips demonstrated that accurate results could be obtained (standard deviation of the difference between the horizontal and vertical strain of the specimen at 11 microstrains during the heating periods). The strain curves also showed high correlation with the temperature curves, thereby indicating reliable strain measurement results with a low uncertainty of 11 microstrains at temperatures up to 700°C.
- In Work Package 4 the hardware and software of the CreepImage system was integrated into a compact prototype system.
Firstly, an initial version of the CreepImage system was built. The opto-mechanical hardware was designed to enable rapid deployment of the system, and a thermal shielding enclosure was designed to protect the hardware from the high-temperatures that will be experienced when the system is used to monitor the creep deformation of an in-service steam pipe in a power plant.
For creep strain measurement during an outage of a power plant, key issues of the system include system positioning and illumination. A laser-based positioning and alignment module was designed to achieve accurate system positioning and optical alignment with accuracy of ±1mm. However, due to the complex nature of this design it was not implemented.
For creep strain measurement at high-temperature, thermal shielding of the CreepImage system is the major challenge. In order to minimise the time the CreepImage system is exposed to the high-temperatures in the locality of power plant steam pipes, the system has been designed to enable quick on-site set up. This is a non-trivial task as accurate creep test measurements necessitate precise positioning and alignment of the inspection equipment.
Secondly, a platform that integrates the hardware and software of the CreepImage system has been developed. This includes the Object Linking and Embedding (OLE) software and an OPC server (OLE software for Process Control) that enables interoperability of the CreepImage equipment and standardization of the control mechanism, a user-friendly human machine interface (HMI) that facilitates image acquisition, camera configuration and image visualisation, and software controls for incorporation of the image correction and DPDIC algorithms.
In order to develop the required remote control functionality in a fast and efficient manner, OPC (OLE for Process Control) software was developed. This approach is widely used by industry to facilitate communication between Microsoft Windows based programs and hardware devices.
The control software operates as a web-based application, connecting an OPC server with an OPC system service to the developed web application (the OPC client). The OPC software accepts various data inputs, translates them into an OPC format, and stores them in a common database. This process enables display and analysis of the CreepImage data. System peripherals (such as the camera module) are connected to the OPC server via the OPC tunnelling
The digital images produced are stored in the common database; the control algorithms retrieve the data from the database and perform calculations based upon the stored images. The results produced are also stored in the database and can be accessed by the user through the HMI. The CreepImage control mechanism operates the image module which comprises a camera, lens and lighting system. Management of the image module is performed remotely via the HMI which is connected to the OPC server. The HMI was designed to provide a user-friendly environment in which an authorised operator can easily connect the camera to the network, configure the camera and manage the captured images/videos. An intuitive user-interface has been designed that enables the operator to quickly become familiar with the required control and management processes. The image correction and DPDIC algorithms were integrated with the HMI using a dynamic link library (DLL). The DLL contains all of the functions necessary for pre-processing captured images (e.g. alignment of images) and to calculate the creep deformation of the specimen under test.
Thirdly, the CreepImage prototype system has been further optimised by incorporating a three-legged positioning mechanism, which enables rapid deployment of the CreepImage system and reliable image acquisition on a length of power plant steam pipe. The new version of the system was tested in the laboratory using a P91 pipe section sample, representative of the type used for high-temperature steam pipework in a power plant. Modifications to the CreepImage system hardware have enhanced both the user experience and the quality of the creep strain results obtained. Laboratory tests and demonstration of the optimised system to project Partners have shown the prototype CreepImage system to deliver the required level of creep strain measurement accuracy (>100µɛ), and to facilitate rapid deployment, in which a creep strain measurements can be performed in less than five minutes..
A protective casing has been developed that ensures the inspection coupon is protected from the harsh power plant environment, throughout the prolonged creep strain monitoring process. Two welding procedures associated with installation of the CreepImage system have been developed to provide power plant operators with the necessary assurances that the CreepImage monitoring process will not adversely affect the steam pipes under inspection.
Since P91 is prone to surface oxidation, direct application of the DIC algorithm will not produce the required level of creep strain accuracy. Consequently, a grid-based FFT method has been developed for strain measurements when the specimen surface has degraded due to prolonged exposure to high-temperatures. However, the accuracy of the strain measurements achieved is also relatively low (approximately 0.1%).
The laser micro cladding parameters has also been optimised to generate and a high-density grid pattern produced on the inspection coupon. High-temperature oven tests using specimens with this fine grid patterns showed improved creep strain measurement accuracy.
- In Work Package 5 a methodology to prognose the remaining life of a component based on surface creep strain measurements was implemented.
Firstly, a literature review of the available prognosis models was performed. The API-MPC Omega method was identified as the most suitable means of performing creep life prognosis with the CreepImage system. This model is based on the idea that the current creep strain rate, along with a brief history of creep strain rates, provides enough information to estimate the past and predict the future creep behaviour of a component. The API-MPC Omega method is included in the American Petroleum Institute Recommended Practice 579 on Fitness-for-Service document, as a reliable method of predicting creep failure in high-temperature applications using commonly used metals. One advantage of this method is that the aforementioned fitness-for-service report includes all of the material constants needed for several commonly used high-temperature application metal alloys. However, it should be noted that the report does not include values for the particular material constants needed for this application.
The concept of Omega as a material property on a par with strain rate draws attention to differences among materials in their response to creep. Some are highly strain tolerant and fail by plastic collapse after large strains, while others lose load-carrying capabilities due to creep degradation after only small strains. If a material is unstable at the microstructure level, due to precipitation, embrittlement, or accelerated softening, such behaviour can be identified from the creep curve. Hence, microstructural issues must be considered when choosing test temperatures. For materials which undergo significant dimensional changes (such as stainless steels), the advisability of applying thermal stabilising heat treatments prior to tests of virgin materials should be determined.
Advantages of the API-MPC Omega method include that it can:
• Predict the time to reach a specified strain.
• Be applied to future service life, independent of past history.
• Provide methods for dealing with tubular components rather than the uniaxial laboratory specimens.
• Allow for relatively short-time benchmark testing at temperatures much closer to operating conditions in the creep range.
• Be incorporated in finite element analysis.
However, the API-MPC Omega method also has its limitations. The integrity of the components that operate at high-temperatures and pressures is dependent on the use of design codes and assessment procedures, that ensure that the components do not fail within the required life time. These codes and procedures utilise the parent material properties of the component, in order to obtain a design or remaining safe life. Moreover, these properties are usually obtained by uniaxial testing on small scale samples of the parent material. The conditions which arise in high-temperature plant, however, differ markedly from those produced by simple uniaxial testing. In reality, components are subjected to multiaxial stresses and, because welding is often used in their fabrication, they contain weld metal and heat affected zones with properties that may differ from those of the parent material. It is therefore important that due consideration is given to the presence of weldments and multiaxial stresses when making creep life predictions
The CreepImage approach (based upon the API-MPC Omega method) has been used to predict the strain behaviour of SS316 virgin specimens and the results have been verified by experimental means. The uncertainty of the predicted rupture time is less than 2%. Hence, it was concluded that the Omega Model can predict the end-of-life of a specimen with high accuracy.
However, comparing the overall results of the API-MPC Omega method with the experimental data gave an error of 17%. A possible reason for this discrepancy is the heating and cooling cycle that was present throughout the accelerated creep test. Since the applied stress (208MPa) was well above the yield stress (205-290MPa at room temperature and around 150MPa at 620°C), repeated plastic strain occurred throughout the test (the initial test run had a plastic strain of 2.3%, and at the end of the test the total plastic strain was around 2.59%). This is likely to have had a significant detrimental effect on the life of the specimen.
Secondly, a creep monitoring protocol was established, which incorporated the CreepImage approach to inspection/creep life prognosis into the current power plant inspection strategies.
Currently, creep assessment is initiated most commonly by the client (as advised by a Technical Service Provider (TSP), or internal engineering department), as part of their asset management and predictive maintenance, or plant replacement strategies, rather than regulatory necessity. The assessment usually takes the form of a three stage process:
• Stage 1 – Desktop assessment of the power plant design and its operational parameters (usually performed at the ‘mid-life’ point of the plant).
• Stage 2 – Site measurements/confirmation of design and operational process to back-up, or confirm the Stage 1 assessment.
• Stage 3 – May involve NDT plant monitoring stress analysis and replication at a later time, unless some components have prematurely reached an advanced stage of creep damage. It should be noted that much of the Stage 3 information will be obtained during the course of statutory shutdown inspection activities, and may include replication and hardness testing.
There are no formal standards or regulations for the assessment of creep damage, or the assessment of the remaining life of a component. The purpose of the scheduled plant outages is to gather sufficient data on the condition of the high-temperature systems to satisfy all parties that the plant can be safely returned to service for a defined period of time. However, the power plant can be returned to service with restrictions on its operation in circumstances where it has not been possible to reliably underwrite the plant’s integrity for the next operating period. In such instances, a safety clause would be required, which may require that the plant is brought off-load, after a short period of time for re-inspection.
It is anticipated that the integration of the CreepImage inspection technique into existing inspection strategies will be a straightforward process. The most complex/time-consuming aspect of the CreepImage approach is the installation of the inspection coupon (that incorporates a micro gird pattern produced using the micro laser cladding process or other technique) and the installation of the protective casing used to protect the coupon from dirt/dust or damage during the period of inspection (typically inspection will be performed over a number of years). The installation of a single coupon and its associated protective case (via suitably qualified welding procedures) is expected to take around one-hour, depending on the accessibility of the inspection area. An initial image/video clip will then be recorded using the specifically designed image acquisition jig. This process will take less than five minutes (including laptop/hardware setup). Once the installation and initial inspection process has been performed (during a planned outage), subsequent inspections can be carried out either during power plant operation or during subsequent outages. Similarly, this simple image/video capture process will be performed in less than five minutes, once access to the inspection area has been achieved. Integration of the CreepImage inspection process into existing inspection strategies will be further simplified by performing the strain calculations for a number of inspection areas, via off-line batch data processing away from the power plant, following the data acquisition procedure.
The types of componet that can be monitored using the CreepImage system has also been defined. The most informative assessment of component condition is obtained during an outage. Components typically assessed for creep damage (operating at temperatures >400°C) include:
• Main steam and hot reheat pipework systems.
• Header systems, manifolds and drains.
• Valves and steam chests.
• A wide range of welds (pipe-to-pipe, large branch, pipe/header penetrations, attachments).
• High pressure and intermediate pressure turbines (rotors, casings, etc.).
• Any component identified as having high stress levels via finite element analysis or other theoretical calculation.
The CreepImage system is capable of inspecting all of the above components provided there is sufficient surface area to attach the inspection coupon, and there is adequate access to the component (the CreepImage optical system requires a 1m stand-off distance from the component under test).
An operating procedure for the CreepImage system that is compatible with current power plant inspection strategies has been formulated. The CreepImage implementation strategy is focused on inspection services during a power plant outage. However, in-service inspection is not excluded, provided that all safety regulations regarding high-temperature working are met. The CreepImage operating procedure is defined as follows:
1. Obtain permission to perform on-site inspection from the relevant authority.
2. Ensure appropriate personal protective equipment (PPE) is worn by all inspection personnel.
3. Erect barriers and signs to prevent unauthorised personnel entering the inspection area.
4. Perform a risk assessment of the working environment and take necessary actions to minimise risk.
5. Initial inspection (performed during a power plant outage):
a) Identify appropriate inspection sites according to the requirements of the power plant operator (in conjunction with theoretical/simulation analyses). Inspection sites are usually situated around welds and incorporate the heat affected zone.
b) Ensure there is adequate access to the component under test.
c) Expose the surface of the component to be inspected (e.g. remove insulation from pipework).
d) Weld an Inconel 625 coupon incorporating the speckle/grid pattern to the area of the component to be inspected.
e) Weld the protective casing to the surface of the component to be inspected (this also ensures that the camera is correctly positioned throughout the entire inspection process).
f) If necessary, apply post weld heat treatment (PWHT) to release weld residual stress.
6. Coupon inspection (performed during an outage, or whilst the power plant is operational):
a) Ensure there is adequate access to the component under test.
b) Expose the surface of the component to be inspected (e.g. remove insulation from pipework). Particular care must be taken with hot components during in-service power plant inspection.
c) Check if the speckle/grid pattern has remained intact.
7. Image acquisition:
a) Connect the camera to the laptop.
b) Mount the high-definition camera (with integral telecentric lens and LED lighting) in the positioning holes in the top of the protective casing.
c) Run the CreepImage image capture software and observe a clear in-focus image of the area under test on the laptop screen.
d) Select the Capture Image option to capture a still image or a video clip of the component under test.
e) Replace any component covering removed prior to inspection (e.g. replace insulation on pipework).
f) Use the CreepImage software to enhance the clarity of the captured speckle pattern and aid image correlation (if required).
8. Subsequent inspections:
a) Run the DIC software (either on-site or after leaving the site) to obtain the accumulated creep strain value.
b) Enter the accumulated creep strain result into the creep life prognosis model to perform a remaining life assessment for the component under test.
c) Export the CreepImage Report using the reporting tool integrated in the CreepImage software.
After an appropriate time interval (6-18 months) repeat steps 1-8 to capture additional speckle pattern images.
The information included in the CreepImage Inspection Report can be divided into three categories:
1. General information on the inspection (input by the user).
2. The CreepImage analysis results (produced by the software).
3. The operator’s comments (input by the service provider).
More specifically, the information included in the report includes:
• Inspection date.
• CreepImage version number (to record any post-project changes to the system specification (e.g. a different hardware setup, DIC algorithm, or methodology).
• Name of the power plant under inspection.
• Name of the service provider and the operator’s name.
• Information relating to the component under inspection:
o Component ID
o Unit
o Weld
o Weld filler
o Component material
o Component/weld service duty
• Results of the CreepImage analysis:
o DPDIC
o Horizontal/vertical/shearing strain
o Strain rate
o Omega value for the creep life prognosis
o In-service/operational time of the component
o Creep life prognosis results
• A ‘free text’ section, where the operator can enter any additional comments (e.g. possible recommendations, or resulting actions).
• One or two digital images of the component under inspection (e.g. photographs or engineering drawings) for identification purposes.
Health and safety issues are of pramount importance in a power plant. Consequently, a Method Statement and Risk Assessment for the prototype CreepImage system have been produced (these two documents typically make up the workplace safety plan). In combination, they provide detailed guidance on how to carry out a CreepImage inspection safely. These documents were specifically developed in order to gain permission from the power plant operator to perform the CreepImage field trials. They must be referenced during a safety induction and then as required throughout the inspection process. The CreepImage Method Statement gives specific instructions on how to safely perform a creep strain measurement using the CreepImage prototype system in an operational power plant. The Risk Assessment determines quantitatively the risks related to the hazards associated with carrying out a CreepImage inspection, as well as the likelihood of them occurring. It ensures that all risks are understood and if they are acceptable.
- In Work Package 6 comprehensive system testing was performed to demonstrate that component life can be predicted with sufficient confidence to minimise the likelihood of catastrophic failure.
Firstly, extensive trials of the CreepImage system were performed including: power plant field testing to assess usability; ambient temperature testing to determined strain measurement stability and repeatability, and a high-temperature creep testing to determine strain measurement capability.
During the on-site preliminary trials of the CreepImage system (performed in a power station during a scheduled outage) two pipe components were examined. The first was a main valve with a complex geometry, which was used to control the flow of a high-temperature/high pressure steam pipe. The major problem encountered inspecting this component was the limited access and the instability of the surrounding floor. In fact, due to the severely restricted access, it was concluded that it would be impossible to deploy the tripod/camera system in a location suitable for reliable image acquisition. The second location chosen for inspection was a temporary wooden platform positioned on top of a scaffold tower. The scaffolding had been erected to facilitate inspection of a length of main steam pipe, which had already had its lagging removed. The CreepImage system (second version prototype, mounted on a tripod) was setup. No mains power was necessary for this trial, as the camera was powered by a USB cable connected to the laptop, the laptop used battery power, and the ambient light was considered sufficient for the camera, hence the LED ring-light was not activated. Prior to the start of the trial inspection, a speckle pattern was spray painted on to a small area of the pipe (this test was designed to have minimal impact on the power plant infrastructure). Then following initiation of the CreepImage image capture software, live images of the speckle pattern were displayed on the laptop screen. The relative motion between the pipe and the camera was clearly seen. To determine the magnitude of the vibration, two video sequences were recorded (with a frame rate of 11fps). Each video clip lasted approximately 30 seconds. Post-processing of the video clips (performed off-site) showed that the typical amplitude of the vibration was approximately 0.2mm. However, the vibration amplitude reached approximately 0.5mm when nearby engineering works involved a sudden impact. Vibrations of this amplitude were shown not to affect the performance of the CreepImage system. The usability of the CreepImage system was evaluated via these tests.
It should be noted that a full CreepImage field trial would have involved the attachment of an inspection coupon and a protective casing (to prevent contamination of the coupon) to the surface of the component under test. However, due to the strict requirement to ensure that inspection process did not interfere with the power plant infrastructure, this was not possible. Laser welding is recognised as the most appropriate technique for welding thin sheet to thick section materials. However, laser welding is not a portable process and therefore it is unsuitable for use in an operational power plant. Consequently, the portable and compact TIG welding process was identified as the joining technique best suited for this application. In order to gain permission for a test that involves welding onto a steam pipe it will be necessary to demonstrate to the power plant operator that the welding operations do not affect the structural integrity of the steam pipe. Hence, qualified welding procedures must be developed to ensure that any residual stresses do not adversely affect the pipework. Two TIG welding procedures (for the attachment of the inspection coupon and the protective casing to a P91 pipe) were developed as part of the project. However, further work is required to fully qualify the procedures, in order that they may be accepted by a power plant operator.
Ambient temperature laboratory trials were performed to demonstrate the stability and repeatability of the CreepImage system. The trials started with the installation of an Inconel 625 inspection coupon and a protective casing on an ex-service P91 pipe specimen. The image capture module was then located in the positioning holes in the top of the protective casing. This ensured that the correct camera orientation and stand-off distance. Once the image capture module was correctly positioned, the image acquisition process was initiated and a series of video clips (lasting approximately 30 seconds) were captured. The video clips were then post-processed to calculate the apparent strain of the coupon. Results showed that although the fluctuation in apparent strain was around ±150microstrain, the mean strain was less than 20microstrain. Hence, high strain measurement accuracy can be achieved by simply averaging multiple measurement results.
High-temperature laboratory trials were also conducted, which validated the strain measurement capability of the CreepImage system on SS316 over a prolonged period. The procedure for the oven creep test was as follows:
1) Fabricate the creep test specimen and produce the required grid pattern on its surface using the micro laser cladding process. The grid pattern consists of a series of evenly distributed small dots of Inconel 625 (dot separation 0.1mm dot diameter 0.1mm).
2) Capture an initial image of the grid pattern using the CreepImage system.
3) Weld three thermocouples on to the side of the specimen (top, middle and bottom positioning). The three thermocouples are used to monitor the temperature distribution across the specimen.
4) Install the specimen and load frame in the creep test oven and apply the loading.
5) Connect the two extensometers to the control computer.
6) Start the test by heating the specimen to the predefined temperature (e.g. 620°C).
7) Once the required temperature is reached, apply the required load (calculated to produce the required tensile stress in the specimen).
8) After a predefined time, stop the test by removing the load and switching-off the oven.
9) Once the oven has cooled down to ambient temperature, move the oven to expose the specimen to the CreepImage camera.
10) Capture additional images of the grid pattern using the CreepImage system. Note, due to access restrictions around the oven the specimen and the loading frame were removed from the oven for the image acquisition process.
11) Go to Step 4 to resume the test (repeat this sequence until the specimen ruptures).
Ideally, the high-temperature creep test would have been performed using an oven with a glass viewing window. With such an oven, digital images could have been acquired throughout the test without interfering with the specimen. However, as this type of oven was not available, a conventional creep test oven was used. The main drawback to this approach was that it was necessary to stop the test and wait for the specimen to cool down before image acquisition could take place. However, an advantage of this approach is that the heating and cooling cycle resembles the temperature variation experienced by a coal fired power plant’s steam pipes, as the plant is brought on and off-line to match varying output demand (there may be around 250 temperature fluctuations in one-year for a typical coal fired power generator). It should be noted that this temperature fluctuation is one of the main reasons why high-temperature paint is unsuitable for producing the required DIC grid/speckle pattern on a power plant’s steam pipework over the long-term.
An accelerated creep test was performed on a SS316 specimen with a grid pattern superimposed on its surface. During the test, the specimen was subjected a tensile stress of 208MPa at 620°C. Three thermocouples and two extensometers were attached to the specimen to monitor the temperature and elongation. The elongation data was used to calculate the creep strain between different test runs, and the results were compared with the strain measurements made by the CreepImage system. A total of 15 test runs (with durations ranging from seven to 94 hours) were performed. Although the entire test had a duration of just 528 hours (at full load and temperature), the actual time to perform the trial was almost two months, due to the need to temporarily halt the creep test (to allow the specimen to cool down to ambient temperature) each time a DIC measurement was performed.
It should be noted that the DIC technique can only measure the total remaining strain of a creep test specimen, and it is not possible to differentiate between creep strain and plastic strain from the total remaining strain. However, it is possible to use the elongation data recorded by the test machine to detect which part of the total strain is associated with creep deformation. This is because creep strain is a time dependent deformation that typically occurs at high-temperatures.
After processing all of the captured images and elongation data, the creep strain and strain rate of the specimen were obtained. The strain rate was high in the primary creep stage, it then gradually decreased, and stabilised at approximately 1.8x10-4/h from 40 to 250h. It then increased from 250h until its final rupture at around 530h.
Since a considerable number of small cracks were found on the specimen surface at 528h, the test was terminated without actually breaking the specimen. This decision was made because it was believed that the specimen was only a few hours away from final rupture (the accumulated creep strain was calculated at 17.71% and the strain rate increased from the steady state rate of 1.8x10-4/h to 6.34x10-4/h). The gauge length of the specimen after Test Run 15 was measured to have an extension of 7.1mm. Since the original gauge length was 35mm, this extension gave an overall tensile strain of 20.3%. The difference of 2.59% between total strain and total creep strain corresponded to the plastic deformation. However, the accumulated total remaining strain from the DIC measurement was 20.11%. This value is very close to the total tensile strain of 20.3% (determined by the gauge length measurements), which means an error of only 1%. In view of the fact that the material beyond the gauge length of the specimen also deformed during the creep test, the overall tensile strain value of 20.3% is likely to be an overestimate. Therefore, it was concluded that the accumulated remaining strain value of 20.11% (obtained from the DIC technique) most accurately represents the actual total deformation of the specimen. This demonstrates the high accuracy for creep strain measurement of the CreepImage system.
Secondly, comprehensive data analysis was performed on the strain measurements obtained using the CreepImage system on a virgin SS316 specimen during an accelerated creep test at 620°C. High correlation (with an R-squared value of 0.9996) was observed between the strain results obtained using the CreepImage system and the creep strain calculated from the elongation data recorded using extensometers. Consequently, it was concluded that the CreepImage system is capable of providing highly accurate creep strain measurements on SS316.
Although it was not possible to perform a complete creep strain measurement in a power plant within the two-year duration of the project, preparations have been made to start a full CreepImage creep strain field trial, during the next scheduled outage at the EDF West Burton Power Station.
Potential Impact:
The final results are a creep strain measurement system and a remaining life prognosis methodology. These include (1) DPDIC algorithm for creep strain measurement; (2) Image correction algorithm to correct distortion caused by high temperature haze; (3) Integrated CreepImage system for creep measurement during power plant outages and whilst a power plant is in operation; (4) Remaining life prognosis methodology based on the API 579 Omega model using creep strain measurements; and (5) Trial test results.
Potential impacts from the project include increased probability of identification and detection of creep damage in power generation (both thermal and nuclear) and petrochemical processing plants. This will decrease the probability of failure of high temperature components due to creep damage and type IV cracking in particular. These types of failures not only result in costly enforced outages, they are frequently accompanied by potential risk to human safety. The optical strain measurement system has wide applicability. In particular gas turbines, steam turbines and steam components could benefit. Hence, the structural integrity of such systems and components will be maintained in order to reduce the likelihood of catastrophic failure.
A series of appropriate dissemination activities were identified and performed by the project Partners. A project website comprising a public area for the dissemination of information, and password protected members area for confidential project files has been created (www.creepimage.eu ). An article that details the aims, approach and anticipated impacts of the CreepImage project has appeared in TWI’s Connect magazine (Issue 185 - July/August 2013). A CreepImage article has also appeared on the website of the UK’s Energy Generation and Supply Knowledge Transfer Network. A conference paper was produced by the project Partners TWI, Dantec Dynamics and Aimen for the British Society for Strain Measurement's (BSSM) 9th International Conference on Advances in Experimental Mechanics, held at Cardiff University, 3rd-5th September 2013. A three-minute promotional video has been produced for public release. It can be viewed on the CreepImage website (www.creepimage.eu/publication) or on the YouTube website at www.youtube.com/watch?v=8jlutNEF2Yg.
To date three major dissemination activities have been performed to promote the CreepImage project. A fourth will be carried out on the 31st October 2013. A presentation was delivered at the EDF West Burton Power Plant in order to gain permission to perform the essential on-site field trials; a conference paper was presented at the 9th International Conference on Advances in Experimental Mechanics; the CreepImage system was promoted at the BINDT NDT 2013 Conference; and the CreepImage system will be demonstrated at the R4i Research for Impact Event.
Patent searches performed during the project term resulted in no results relating to high temperature creep strain measurement via DIC. However, Chinese Patent CN 201795989U registered the idea of attaching a coupon to a base material for the purpose of creep strain measurement. It covers the concept of a protective casing and digital image analysis, but not the use of a grid pattern, nor the use of a telecentric lens for image acquisition to minimise the apparent strain which is inevitable if a conventional lens were used to inspect a curved surface on a high temperature component. It is considered that this patented idea cannot achieve accurate creep strain measurements claimed by the patent owners. Moreover, extensive literature searches have found no evidence that the patented ideas have been practically tested.
The patent searches and literature surveys performed during the project have demonstrated that the CreepImage SME Partners could potentially submit a number of patents relating to the advancements made in the CreepImage project.
Possible patent applications could cover:
• High temperature strain measurement algorithms.
• Distortion correction algorithms.
• Pattern production methods for optical measurements.
• The entire CreepImage system (assembled components).
However, at the present time there are no plans for the SME Partners to apply for a patent, as Partner discussions have identified the need to further investigate possible markets and to arrange more extensive field trials to fully validate the inspection technique. It should be noted that a full creep strain monitoring trial (performed in an operational power plant) would have a minimum duration of several years, due to the prolonged nature of creep damage formation and the lengthy interval between successive power plant outages (typically two years for a minor outage and four years for a major outage). Nevertheless, the SME Partners will still be in a position to benefit from the CreepImage developments through their increased know-how of image analysis, DIC techniques and remaining life prognosis methodologies.
Overall Strategy for Exploitation
The exploitation and commercialisation of the CreepImage system will occur in several distinct phases, covering product development, initial exploitation, European expansion and global expansion. The exploitation strategy following project closure is outlined below.
Phase 1: Demonstration of the capability of the CreepImage system to predict the remaining life of pipework during power plant outages (years 1-4)
1. The first demonstration of the CreepImage system performing creep strain measurements to potential customers will be initiated by installing the inspection coupon and associated protective casing on a steam pipe during a scheduled outage of a power plant. A subsequent scheduled outage is then required to obtain a second image of the coupon at a similar ambient temperature. In this way, the creep strain accumulated between the two outages will be measured and the creep strain rate calculated. In addition to the measurements made during outages, measurements in an operational power plant will be performed following the first outage. However this process will be significantly more complex, due to the additional risks involved with working close to high-temperature steam pipes.
2. Comparison between conventional creep strain and CreepImage methods.
3. Improvements to the CreepImage prototype.
4. Improvements to the CreepImage operating procedures.
Phase 2: Development of an Industrial System (years 1-2)
1. Ensure the prototype is physically robust and suitable for the harsh environment of a power plant.
2. Development of a user-friendly and easy to use software interface (suitable for a technician to operate).
3. Refinement of the operating procedures, so that they can be easily adopted/integrated into the power plant operator’s working practices.
Phase 3: Demonstration of the capability of the CreepImage system to predict the remaining life of pipework in an operational power plant (years 2-7)
1. Development of methods to perform creep strain measurements on high-temperature components.
2. Development of a heat protection system for the CreepImage equipment.
3. Development of operating procedures for high-temperature environments.
4. Further develop the CreepImage creep life prognosis model for high-temperature components.
Phase 4: Industrialisation of a High Temperature System (years 6-7)
1. Making the Phase 2 system sufficiently robust for the high temperature environment of a power plant.
2. The implementation of any necessary software updates.
3. The creation of operating procedures that can be easily adopted by the power plant operators/inspection providers.
List of Websites:
www.creepimage.eu
Contact details of the project's coordinator:
Jianxin Gao
TWI Limited,
Granta Park, Great Abington
Cambridge, CB21 6AL
UK
Tel: +44 1223 899000
Fax: +44 1223 892588
E-mail: jianxin.gao@twi.co.uk