PIPES FOR LIFE - AN NDE INSPECTION OF AGR BOILER TUBES

C. R Bird, D. Maclennan, Doosan Babcock, UK

Abstract As part of the Lifetime Extension Programme for gas cooled reactor power stations EDF Energy has embarked on a detailed study of the condition of their boiler tubes. As part of this study Doosan Babcock were requested to develop an inspection system for the measurement of the wall thickness and bore diameter of the superheater boiler tubes.

An ultrasonic head has been designed to negotiate pipes with a 13mm internal bore after first negotiating 5 off 45 degree bends. The requested probe deployment distance was approximately 30m. To add a further challenge there is a restriction which forces the maximum diameter of any probe to approximately 11mm.

The tubes are finned on the external surface adding to the complication of generating a wall thickness measurement.

This paper describes the technical challenges, the inspection solution and the application of the developed techniques in the field. Keywords: Boiler Tube Inspection, Signal Processing, Restricted Access.

1. INTRODUCTION The aim of the NDT development and application has been to provide boiler tube wall thickness measurements to enable a quantitative assessment of the boiler tube condition forming part of the boiler lifetime monitoring programme.

The boilers were manufactured in 1976, they were not designed with in service inspection as a consideration, resulting in consequential access and inspection challenges. The boilers layout is shown in Figure 1a.

This paper highlights the extreme challenges associated with the inspection problem and provides an outline of the solutions developed to solve these challenges.

The challenges can be divided into mechanical and NDT. Both the mechanical and NDT challenges were at the boundaries of current best practise forming an intriguing development project. This paper first discusses the mechanical challenges followed by the NDT challenges. The paper then goes on to describe our solutions to the technical challenges. 2. INSPECTION CHALLENGES 2.1 Mechanical challenges. As introduced these boilers were not designed with inspection in mind, resulting in difficult NDT access. The boiler design utilises a compact and complex helical tube pattern, to achieve high thermal efficiency. Figure 1a presents an overall drawing of the boiler.

The only feasible NDT access point for tube inspection of the superheater outlet headers is the top of the boiler. To provide access to the tubes the header dome is cut off and then reinstated after the inspection. The probe is lowered 10m (vertical) through a tail pipe, containing a number of 45° bends to a bifurcation (BIF) where two superheater tubes feed into the tailpipe. (Figure 1b).

Figure 2 provides a section through the BIF. The vertical pipe has an internal diameter of approximately 20mm.

762

Mor

e In

fo a

t Ope

n A

cces

s D

atab

ase

ww

w.n

dt.n

et/?

id=

1855

5

The two smaller pipes that feed the tailpipe have an initial bore diameter of 13.8mm. The forging has no blend radius and a bend angle of 80°, additionally any probe has to identify and navigate around either the top or bottom superheater tube.

Figure 1a Drawing of boiler layout Figure 1b Photograph of the boiler tube bifurcation (BIF).

The superheater tube forms a helical spiral is ~100m and undergoes material and dimensional changes throughout the height of the boiler. The first of theses obstacles is a transition joint approximately 11m upstream of the BIF where the material changes from 316 stainless steel with a 13.8mm bore, to 9%Cr, 1%Mo carbon steel with a bore of 12.7mm.

Eddy-current probes are frequently used for PWR steam generator inspection but they are only subject to 1 or 2 small radius bends. After 360° of bends the capstan effect takes hold and prevents deployment via pushing or pulling any cable.

An access solution was required. Self propulsion systems are regularly used e.g. “PIGS”, but these normally have pipe diameters of many centimetres (50mm minimum) and are not subject to bends where the diameter of the pipe equals the bend radius to be navigated.

Ultrasonics inspection was chosen as the NDT method to provide wall thickness measurement. Given the dimensional limitations imposed by the requirement to pass around the BIF, it was decided that an immersion technique was the only feasible solution given the mechanical complexity of providing a contact probe. A further advantage of an immersion solution was the possibility of providing bore diameter measurement.

10m

Superheater headers

BIF Joint

763

Figure 2. Section through BIF

The consequence of using water immersion was that the probe would be small relative to the

bore of the tube. It was important that the probe could be held central within the tube to ensure the beam was near perpendicular to the surface. A flexible centralisation system was required to ensure high accuracy measurements.

In common with all boiler circuits there are strict Foreign Material Exclusions. In short what goes into the tube must come out. Being a nuclear installation these rules are strictly adhered to and the probe design was required to demonstrate that it would not break and even more importantly would come out after deployment. Furthermore all materials used in the construction of the probe are required to be nuclear compatible.

2.2 NDT Challenges The NDT challenges can be divided into four areas:

i) Tube dimensions and length ii) Probe alignment iii) Finned tube iv) Probe size

20mm

13.8mm

764

2.2.1 Tube Dimensions and Length. The inspection technique was required to measure wall thickness reliably with accuracy better than 0.2mm. Whilst this is not difficult on flat plate with a good surface finish where the probe can be manoeuvred to be at normal incidence to the surface the task is far more challenging when the probe is 35m away from the instrument down a 13.8mm tube after navigating an 80° bend with no radius.

The as built tube wall thickness ranged from 1.8mm to 2.4mm depending upon the location within the boiler. This thickness of tube would normally require a high frequency probe (greater than 10MHz and if possible 20MHz) to provide accurate thickness measurements.

The cable length and diameter causes high frequency attenuation lowering the possible probe frequency. To date the maximum centre frequency transmitted by this length of cable has been 8MHz. It should be noted that to achieve the required flexibility in the cable and to negotiate the BIF 40AWG, coaxial wire has been used for the cable construction.

In order to negotiate the very small radius bends, see Figure 2 the space envelope for the probe must have a smaller diameter than ideal (6mm at the transducer) and must also be relatively short (20mm in axial direction). Further, the cable was limited to be less than 4mm diameter including the outer tri-axial sheath. When taking measurements this small probe then requires to be centralisation and aligned axially.

2.2.2 Probe Alignment To obtain accurate bore diameter and wall thickness measurements the ultrasonic beams need to be near normal to the internal surface of the tube. The small tube radius creates a condition where a very small probe eccentricity causes a high degree of radial beam misalignment for example 1mm eccentricity causes a 17° radial beam misalignment. Poor probe alignment causes pulse shape distortion and reduction in measurement accuracy.

As introduced the probe is required to navigate around a 80° bend and then down a spiral tube of two different diameters. To navigate around the 80° bend a dog bone, (Figures 2 and 5), shaped probe was designed and manufactured incorporating centralisation features. It was recognised at the technique design stage that alignment to the surface within 2.5° would not be achieved in all occasions, and that it would be necessary to use the ultrasonic data to discriminate between aligned and non aligned ultrasonic beams.

The dominant ultrasonic energy is received from the beam path which is at normal incidence to the bore of the tube, represented in Figure 3 as a solid line. The dashed line represents the normal beam direction from the transducer face. In this case the 6dB half angle beam spread is approximately 2.5°. Hence the algorithm used to measure the bore diameter determines a best fit circle taking into account the three beam paths normal to the inner surface of the tube.

Figure 3. Exaggerated illustration of measurement accuracy challenge due to radial alignment

765

In addition to probe concentricity, axial misalignment must also be taken into account. The energy normal to the tube bore will dominate but when the beam is non normal to the tube bore the ultrasonic signal becomes extended, this leads to difficulty discriminating the correct signal peak.

2.2.3 Finned Tube Wall Thickness Measurement The majority of the tubes to be inspected were manufactured with fins as shown in Figure 1b. The fins are manufactured with rounded “roots” causing the signals from the outer surface (at the base of the fin) to be severely reduced and scattered by the radius. Furthermore the beam will detect the fin root even if it is not perpendicular in the radial plane. Figure 4 provides a picture of a sectioned tube and illustrates the thickness measurement challenge. Incorrect alignment will cause a distortion in the pulse shape and a potential inaccurate beam path.

The finned tube was manufactured by a rolling process similar to bolt manufacture, ensuring that the fins have a consistent and accurate pitch, (2.5mm), and shape. It is possible to take advantage of this regular pattern in the signal processing as discussed in Section 4.

Figure 4. Section through fined tube.

2.2.4 Probe Size To achieve larger beam spread and greater tolerance to beam alignment, small crystals are required (typically <2mm diameter) but for transmission of power and matching crystal impedance to that of the cable a larger crystal is required. The probe case is currently too small to fit a matching amplifier circuit.

3. INSPECTION SOLUTIONS The probe consists of 6 off pulse echo elements forming three pair spaced at 120° intervals. The probes are not twin probes but two rows of pulse echo elements. The two rows are used to measure probe tilt in addition to tube diameter and thickness. Measurement of probe tilt and the concentricity of the probe is an essential parameter in assessing the accuracy of the measurements achieved and selection of the appropriate signals.

Correct beam position

Incorrect beam position

766

Figure 5. Photograph of deployed probe showing centralisation features.

Probe deployment is via a mixed air/water deployment system and the data is collected during probe withdrawal via a patent water propulsion system. Beads are fitted along the length of the probe to provide propulsion during deployment and retrieval. The probe is mounted into a spring section to enable a high level of mechanical flexibility to aid the deployment around the 80° Bifurcation port. Figure 5 illustrates the physical size of the probe and the centralisation features used to align the probe once deployed.

Figure 6 provides data from a probe in a finned test piece. The water to steel interface generates a large signal but the OD generates as small signal due to the loss of energy at the interface. In this case we have to detect the signal from the finned root which generates an even weaker scattered signal. Due to the signal scattering conventional repeat OD signals cannot be use for the thickness measurement normal thickness measurement. To accommodate this dynamic range a 12 bit 100MHz digitisation instrument was selected. Signal processing algorithms were developed to facilitate data interpretation.

4. SIGNAL PROCESSING AND DATA PRESENTATION

The signal processing is performed in three stages.

i) Water to steel interface echo detection and straightening.

ii) Water path determination and bore diameter computation

iii) Wall thickness determination.

ID Signals

Weak OD Signals

Repeat water path signal

Row A

Row B

767

4.1 Water to steel interface detection and straightening Figure 6 shows A-scan and B-scan plots for one row of beams, (the three zero degree beams at the 120° positions), from a plain calibration pipe. The plot shows the water to pipe internal diameter (ID) signal and then the small amplitude pipe outside diameter (OD). Further it shows that the water path is not equal in each direction.

Correct peak detection in the data is essential but eccentric probes will cause a distortion in pulse shape and incorrect thickness determination. To aid the processing the data processing straightens the ID interface signal with respect to axial position. This effectively lines up the OD signal and eases the correct identification of the fin root signal. The software includes a number tools to simulate manual data interpretation. The process identifies the ID and OD signal over a length of tube, to generate an accurate mean position of the two interfaces. It is pointed out at this point that the data analysis is designed to identify general trends in the wall thickness (as opposed to corrosion pits or other defect types). An example of some date after this processing is provided in Figure 7.

Figure 7 Image of Heterodyned data and interface echo straightened data.

4.2 Water Path and bore diameter determination After the data is straightened the water path is measured and a best fit circle approach is used to determine the tube bore diameter. This calculation takes into consideration the skewing of the beams due to non–normal propagation as discussed section 2.2.2.

4.3 Wall Thickness Determination The signal processing for wall thickness employs four methods for plain tube and two methods for finned tube. Plain tube processing can take into account repeat back wall reflections to provide very accurate wall thicknesses. Finned tube as already stated, only uses the water to steel interface to finned root path. To this end a calibration scan is performed in a test tube with a known bore diameter to determine the system delay for each element.

768

Figure 8 Screen shot of data analysis after plotting the wall thickness for the three probes

Figure 9 shows data from a finned test piece with a photograph of the sectioned tube above. This finned test tube has five wall thicknesses with a constant bore diameter. The data is presented as a B-scan view, the horizontal axis is encoded distance along pipe. The data shows a series of fin root signals with a selected A-scan at the central position of a fin. To ensure that data is captured from the centre of the fin roots data is captured every 0.5mm along the tube bore both in test pieces and on the plant. Automated data analysis poses two challenges: firstly identification of the central root fin signal and secondly, identification of the correct point on the wave form to provide acceptable wall thickness and diameter measurement.

As introduced the tube is finned but has a regular pitch. This feature is used by the pattern recognition system to identity the centre of the fin root. The signal processing method used is called heterodyning. Once the fin root signal is identified from each root the remaining data is discarded and the data is re-plotted as if the OD signal is continuous. The data is then processed using peak detection software to provide wall thickness measurements.

769

Figure 9 Screen shot of the ultrasonic responses a finned test tube with different wall thicknesses.

4.4 Data Presentation The software outputs the mode, mean and standard deviation for the bore diameter and tube thickness for the section of pipe being analysed. This output is shown in Figure 10.

Figure 10 Statistical output of analysis software.

Second water path signal

ID Signal

Fin root signals

A-scan at a maximum fin root signal position

ID Signal

Fin root signal

770

5. SYSTEM DEPLOYMENT

The inspection system has been deployed under strict trial conditions at two power stations. This has allowed identification of operating constraints and verification of mechanical reliability. The next stage is to assess the reliability and accuracy of the bore diameter and thickness measurements. 6. Conclusions

1 A tube wall thickness and bore diameter system has been developed and deployed down 30m of spiral tube and negotiated 80° bends.

2 A data analysis system and software has been developed and successfully deployed to account for probe misalignment whilst maintaining accurate measurements.

3 The system has proved to be mechanically reliable and is now undergoing the early stages of qualification.

Acknowledgements

1 The authors wishes to acknowledge EDF Energy for their support in this development and application project.

2 The authors acknowledges the advanced signal processing support provided by SNAPE signal processing.

3 The authors of this paper wish to acknowledge the assistance of AMEC towards the mechanical deployment solution.

771