ULTRASONIC PROVE UP IN SUPPORT OF MAGNETIC FLUX LEAKAGE TANK FLOOR INSPECTIONS

INTRODUCTION

It is generally recognized that Magnet Flux Leakage is only a truly qualitative technique and that indications detected using MFL must be assessed using a more quantitative method. Invariably this task falls to Ultrasonics for the accurate assessment of indication severity.

Over the last ten years the author has witnessed many ultrasonic methods and techniques for the evaluation of corrosion in support of MFL Tank Floor Inspections ranging from the sublime to the ridiculous and from the appalling to the ingenious. A very large percentage of the methods witnessed were severely flawed and it was apparent from questioning the technicians involved that they did not have a real understanding of what was required. There is a very big difference between "Thickness Measurement" and "Corrosion Assessment".

Localized pitting corrosion is sometimes very difficult to find and measure using compression wave Ultrasonics techniques due to the very shape of the defect. Many excellent MFL inspections have been compromised by the inability of a bad Ultrasonic method to both find and accurately assess the severity of indications. In most cases it is not the technicians fault as there is very little published information in this regard and even less training available. This is what has prompted the author to publish this paper in the hope that Ultrasonic prove up methods will become more standardized and reliable within the industry. To this end the author will describe below the equipment and method that has, in his experience, proven to give the best results possible in the assessment of MFL indications using the compression wave Ultrasonic Technique.

>It is assumed that the reader has a basic understanding of Ultrasonic principles in order to obtain the most benefit from this paper.

EQUIPMENT

ULTRASONIC INSTRUMENTS

Digital thickness gauges are not ideally suited for this application especially when evaluating localized pitting type corrosion. Most of these gauges use peak or leading edge data from the first and second back wall response to determine thickness. It is almost impossible to obtain two reliable back wall responses from an irregular shaped indication. In the case of very small pits the normal back wall response will obviously interfere with the accurate measurement of the pit depth. The trigger gates are often set relatively high with this type of equipment to prevent false measurements from noise. Indication responses can be very low amplitude and therefore rejected using this method. Scanning to locate small indications is impossible with digital thickness gauges. The technician is limited to a move and measure technique (pecking) that has a limited chance of actually finding the indication at all. This type of equipment is seriously limited in this application and is not recommended.

There are digital thickness gauges that have a rudimentary A-scan display. These are normally small, low resolution displays. Even if the processing speed is up to processing the data real time it is often the case that the screen update speed is too slow to allow any scanning technique to be used. This type of equipment is also seriously limited in this application and is not recommended.

B-scan equipment uses the ultrasonic data to provide a profile view of wall thickness and can be useful as a reporting function but in most cases does not allow the technician access to the A-scan information necessary to carry out a reliable evaluation of the Ultrasonic data. The data for the display is obtained in much the same way as in Digital Thickness Gauges and is therefore seriously limited in this application and is not recommended.

In the opinion of the author the only Ultrasonic apparatus that should be used for the assessment of corrosion indications is one that has a high resolution real time A-scan display that will allow scanning speeds of at least 6 inches per second. Most of the current commercially available instruments are more than capable of meeting these requirements. The very early digital instruments had serious limitations regarding processor and display update speeds. A good analogue display is still superior to most digital displays for this application. [If you can still find one.] Having said that, the latest digital display instruments are more than fast enough to give acceptable results. The A-scan display allows the technician to maximize the response and identify the first and closest facet of the indication for the accurate measurement of remaining wall thickness.

There are differences between instruments from different manufacturers and the combination of transducer and instrument is extremely important. All like configuration transducers from different manufacturers do not always perform the same and should be carefully matched to the ultrasonic instrument being used. A combination of transducer and instrument from the same manufacturer is recommended as a starting point. If the instrument allows adjustment of pulse amplitude and width it is possible to experiment with other transducers to obtain the best results possible.

ULTRASONIC TRANSDUCERS

In this application where nominal plate thickness is in the range of 0.25" – 0.50" it has been determined that, in most instances, the best results can be achieved by using a 5 MHz, dual element transducer 0.375" in diameter. In some cases it is necessary to resort to using smaller higher frequency transducers (surface condition) or larger lower frequency transducers (thicker plates). The implications as regards beam spread and resolution must be considered when using these alternative configurations. Slight focusing of the elements is required to minimize inaccuracy on thinner materials. The chosen transducer should exhibit minimal cross talk even when worn down to its limits. Different transducers vary widely in this regard. A fairly hard wear face is important unless you want to spend a small fortune on replacements. Tank floor plates can be extremely abrasive. Good quality cables are a must.

ULTRASONIC COUPLANTS

There are many proprietary couplants on the market that perform very well in this application. As a general rule the higher the viscosity the better. The author has found that in most cases water and plenty of it is more than capable of achieving the required results and is a lot cheaper.

MARKING MATERIALS

The choice of marking material can have a significant adverse effect on the ultrasonic prove-up. Wax crayons are the worst followed by paint markers. It is sometimes impossible to couple sound through these materials. This is obviously of concern as the mark should be exactly where the indication is located. A good quality chalk is recommended for indication location marking. After Ultrasonic prove up the indications can be marked more permanently with a wax crayon or a paint marker.

METHOD

The method and technique described below is geared towards the assessment of localized pitting corrosion rather than lake type corrosion. By its very nature lake type corrosion is much easier to locate and assess with a high degree of reliability unlike pitting corrosion, which can be extremely difficult to locate and measure unless the correct method and techniques are used.

INSTRUMENT CALIBRATION

Using a step wedge and the transducer of choice, adjust the range and delay controls to display a calibrated range so that the first two back wall responses can be displayed on the screen. In the case of 0.25" nominal thickness plate this would mean two responses at 0.25" and 0.50" respectively. Adjust the gain to raise the first back wall response to full screen height. (Fig. 1.) All responses should be adjusted to the time base. All gates should be disabled at this time. If the instrument has a reject control, make sure it is turned off or at zero. If provided, adjustment of any pulse amplitude or pulse width controls can be used to optimize the shape of the signal envelope.

Lock all controls other than gain to prevent any inadvertent adjustment to the base line setup. Evaluation of the located pit can now begin. Liberally spread the couplant of choice over the area of plate concerned. Place the transducer adjacent to the location of the pit and obtain a back wall response from the nominal plate thickness. The gain should be adjusted, at this time, to bring the first back wall response to full screen height. Note where the first facet (left hand edge) of the first back wall response breaks the time base. This is your nominal plate thickness. Add a further 6db for scanning and note this setting.

Slowly scan the marked location of the pit. Even at this gain setting there may be no discreet signal from the corrosion pit at this time. The only indication of anything amiss may be a dip in the amplitude of the back wall response. Home in on the dip in back wall response and increase the gain until the response from the corrosion pit is evident. (Fig. 2.) Maximize the earliest response from the corrosion pit by small scanning movements of the transducer. Continue increasing the gain until the response from the pit reaches full screen height and note where the first facet of the indication (left hand edge) breaks the time base. (Fig. 3.) This will be your remaining wall thickness measurement to be recorded. If the gain is so high at this point that electronic noise from the instrument is evident, raise the signal response only as high as necessary to discriminate between signal facets and the electronic noise.

Do not forget to adjust the gain to the scan setting established earlier before moving to the next indication. Calibration should be checked at regular intervals as any wear in the face of the transducer will affect the accuracy of the measurements taken. At higher gains electronic noise and cross talk may be evident on the time base but it is relatively easy to differentiate between the noise and a true signal response. It is sometimes necessary to extrapolate exactly where the signal breaks the time base based on the slope of the first facet. (left hand edge)

Using this method it is possible to obtain the most accurate measurement of remaining wall thickness from the Ultrasonic data presented.

RECORDING OF DATA

It is quite common these days to see thickness data presented to three decimal places. In all honesty the author questions the accuracy of the third digit, bearing in mind the small errors that can occur in the calibration routine. In the above described method it is possible to make a judgment as to indications that break the base line between graticules. In this case the author tends to split the reading rather than go to the third decimal. ie. If a signal breaks the baseline between 0.16 and 0.17 it will be recorded as 0.165.

For each indication the following minimum information should be recorded:

Plate number

Indication number

Nominal plate thickness

Minimum Remaining wall thickness (at location of pit)

Measurement datum point (eg. South West corner of plate)

X – Y location measurement and direction from datum (eg. 25" North 16"West)

This will allow the technician or tank owner to accurately locate any indication at a later date should any marking be erased.

GENERAL INFORMATION

Many signals generated during an MFL Inspection may not be due to corrosion pitting. Weld scars, arc strikes and other magnetic anomalies, even lamina inclusions can trigger responses, but whatever the cause it must be identified. If there is definitely no Ultrasonic evidence of corrosion these other causes must be considered and investigated.

Accurate measurement of pitting corrosion is problematical when coatings are present. The thicker the coating the more difficult it becomes. Even if it is possible to transmit sound through the coating material and evaluate the carbon steel plate underneath, accurate pit depth measurement is unreliable because the velocity in the coating is generally unknown and as mentioned earlier it is very difficult to obtain repeat responses from small corrosion pits to allow a peak or edge measurement technique between responses. Invariably the residue of energy being returned from the nominal thickness around the pit will interfere with this technique. Any high amplitude localized signals will require the removal of the coating to properly evaluate the accurate depth of the indication. Low frequency eddy current techniques have been developed to overcome this perceived weakness of MFL and to avoid the necessity of removing coating. The accuracy of the information from such devices is questionable and is dependant on the coating having an even thickness. This is rarely the case on fiberglass coated floors as the coating thickness can vary significantly throughout the floor area.

The ultrasonic energy reflected from a clean back wall will have a relatively tight signal envelope as it is being reflected back from a flat surface. This is not true of the energy reflected from corrosion. It’s irregular profile means that energy is being reflected from many different surfaces and therefore the signal envelope will be much wider on the time base and of a much lower amplitude. This is why it is so important to make significant gain increases to properly evaluate the significance of the indication. Another point worth bearing in mind is that with conical shaped reflectors that have a peak there will be very little, and sometimes no energy reflected from the tip of the indication. In this case it will be seen that the ultrasonic data will underestimate the severity of the indication.[Food for thought!]

The author does not wish to publish an endorsement of any particular manufacturers equipment but would be willing to discuss his experience in this regard on a one to one basis and welcomes any comments regarding the points raised in the above text.

FIG.1

Figure 1 shows ascreen trace depicting a calibrated time base of 0.00" to 0.050". The first facet (left hand edge) of the first back wall response can be seen to break the time base at 0.25". The first facet (left hand edge) of the second back wall response can be seen at the right hand side of the screen braking the timebase at 0.50".

FIG.2

Figure 2 shows the screen trace with the transducer over the top of the corrosion pit. Note the reduction of the back wall response despite the 6db increase in gain. The response from the pit can clearly be seen emerging from the time base at 0.165". At this point a further increase in gain is required to determine the true remaining wall thickness.

FIG.3

Figure 3 shows the trace with the transducer directly over the corrosion pit and the gain increased to bring the signal almost to full screen height. It will be noted that cross talk is evident at this high gain level and it is necessary to extrapolate where the signal breaks the time base from the slope of the leading edge. In this case it would be a reasonable assumption that the remaining wall thickness is in the order of 0.150".

FIG.4

Figure 4 shows the trace with a gated threshold and digital read out in operation. It can be seen that the elevated time base represented by the gate can produce an error in the order of 0.020". Even if the gain is increased to that in figure 3 it will be seen that a significant error would still exist. NOT RECOMMENDED.