ULTRASONIC PIPE INSPECTION PROVE-UPS CAN BE IMPROVED

B.J. Hodges
Lone Star Steel
Lone Star, Tex.

In extensive tests conducted in its laboratories, Lone Star Steel found conventional manual ultrasonic prove-ups of pipe imperfections to be consistently inaccurate.

Error rates ranged from 65% to 100%. On an average, we found that of every five joints of casing rejected by manual prove-ups, four did not have rejectable imperfections.

In many cases, the rejected pipe cost several thousand dollars a joint. The cost of rejecting acceptable pipe is not one we or any other pipe manufacturer can simply absorb. Buyers eventually pay the cost by paying higher overall prices.

Yet, our research shows unnecessary rejection is a cost that can be easily eliminated by making simple changes in conventional ultrasonic prove-up procedures.

Our proposed changes do not involve loosening existing standards. But we are convinced that it is to our and the oil and gas industry's best interests to ensure prove-up procedures produce accurate and repeatable results.

CURRENT INSPECTION PRACTICES

Normally, pipe is inspected by automated or semi-automated ultrasonic or electromagnetic non-destructive testing equipment. The equipment is calibrated to sensitivity levels that will locate imperfections smaller than the maximum size allowed by the applicable specifications.

Each potential imperfection is then manually proved-up.

If the imperfection is on the outside surface, the surface can be filed or ground. Magnetic particle inspection techniques then can be used to locate the depth of the imperfection.

If the imperfection is on the inside surface or beneath the surface, manual ultrasonic prove-up is used to determine the depth of the imperfection.

Accepted practice has been to use a 1/4-in., 2.25-to-5.0-Mhz transducer. It is mounted on a plastic wedge that has been hand contoured to the pipe's surface to hold the transducer so the sound beam will enter the pipe at a 45 angle and perpendicular to the pipe's centerline.

The unit is calibrated to a notch cut to 10% (API N10) or 5% (API N5) of the pipe's wall thickness. The notch is cut in pipe of the same diameter and wall thickness as the pipe to be inspected. Specifications indicate the notch should have parallel sides, to be directed to the pipe's centerline, and to have a depth that doesn't vary more than 15% from specified depth.

INDICATIONS OF A PROBLEM

Lone Star decided to carefully examine sixty imperfections that had caused rejection of 5 1/2-in. OD by 0.304-in. wall thickness pipe. The inspector had used a 1/4-in., 3.5-Mhz transducer calibrated to an API N10 notch depth (0.0345-in./O.0255-in. for this pipe).

Prove-ups on the same imperfections were performed with a 1/4-in., 2.25-Mhz transducer calibrated to the same standard.

We then prepared polished micro specimens from each section of the pipe where rejectable imperfections had been located and physically measured the actual depth of the imperfection.

Imperfections classified as rejectable by the 3.5-Mhz transducer used by the third party inspector actually ranged in depth from 0.00 in. to 0.040 in. In fact, 85.3% of the defects recorded as rejectable using the 3.5-Mhz transducer were well within acceptable specifications.

Our 2.25-Mhz transducer didn't do much better. It had a surprising, extremely high error rate of 65%. To see if the problem was peculiar to this particular lot, we immediately began micro analyzing prove-up rejections from all pipe sizes, thickness and grades.

We found error rates continuously ranged between 65 and 100%.

THE VARIABLE CALIBRATION NOTCH

The discovery of general inaccuracy led investigations of all areas of ultrasonic prove-up inspections to see if the cause of the high error rates could be identified. We first looked at calibration notches.

Specifications allow a 15% tolerance for notch depth, so variations can be appreciable. Compounding the problem, calibration notches usually are cut in the field using a smaller diameter rotary saw blade. We found field-cut notches were consistently less than the nominal specified depth, even when allowing for the 15% tolerance.

Instruments were being calibrated to reject pipe for imperfections that were well within specifications.

Further compounding the problem, the sides of the field-cut notches, instead of parallel, were more likely to be V-shaped or buttress or dovetail notches. As a result, signal response from one side of the notch to the other could vary as much as 50%. To compensate, the ultrasonic inspection equipment is calibrated to average the response of the two sides or to the most critical of the two sides.

In effect, the calibration was more a function of the orientation of the notch's sides than of notch depth.

To eliminate those variables in our own operations, we specially prepared calibration notches for each of our non-destructive testing areas. The 2-in. long notches were cut by an electrostatic discharge machine because it offered exceptional accuracy in depth penetration, excellent parallelism of notch sides, and presented a surface that was more representative of an imperfection surface than notches created by other methods.

Our calibration notches are required to be certified to be within 0.002-in. of nominal depth and the prove-up signal response from one side of the notch to the other must be within 1 db (approximately 10%). The prove-up unit is calibrated against the critical side of the notch. The signal response is adjusted to 90% of screen height.

In our prove-ups, any response of less than 80% of screen height represents an imperfection that is well within specification limits.

THE VARIABLE WEDGE

A plastic wedge is used to hold the transducer so the sound beam will enter the pipe at a 45 angle, the ideal sound beam to pipe refraction angle. Usually the operator contours the wedge by rubbing it along sandpaper mounted upside down on the pipe. In addition, the wedge wears and is frequently recontoured.

As a result, the refraction angle constantly varies during prove-up operations.

For our own prove-up operations, we now mechanically contour the wedge to the average of the nominal and maximum pipe diameter in each size. We also put wear studs on each corner of the wedge to maintain a more constant refraction angle.

THE WRONG TRANSDUCER

Because of the high error rates encountered with the 3.5 Mhz, 1/4-in. transducer, we conducted tests to evaluate the accuracy of 2.25 Mhz, 1/4-in. transducers. We placed 12 notches on the inside surface of a 13 5/8-in. OD by 0.625-in. wall thickness pipe ring. Notch depths ranged from 0.063 in. to 0.007 in.

We calibrated the oscilloscope screen to show a spike at 100% of screen height on the 0.063-in. notch.

Though readings on the next two notches should have been less, they too were 100% of screen height. Defects within acceptable limits were being recorded as rejectable.

We saw similar results on a 16-in. O.D. by 0.495-in. wall thickness ring, a 9 5/8-in. O.D. by 0.395-in. wall thickness ring, and a 5 1/2-in. O.D. by 0.304-in. wall thickness ring. In all, we tested 36 notches. Of those, only four should have given us a 100% screen height response. But, as shown in Figs. 1, 2, and 3, we saw a 100% response from 14 notches, a 71% error rate.

These results, along with the results from approximately 200 examinations of polished micro specimens made it clear that the '/4-in. transducer is inadequate for ultrasonic prove-up of imperfections in steel pipe with a wall thickness of 0.300-in. and greater.

THE ADVANTAGE OF 1/2-IN. TRANSDUCERS

We conducted the same tests using a 1/2-in., 2.25 Mhz transducer. As you can see in the curve for the 1/2-in. transducers in Figs. 1, 2, and 3, the error rate was zero.

We then tested the 1/4-in. and 1/2-in. transducers on imperfections ranging from cracks to surface scars.

With the 1/4-in. transducer, the error rate from surface scars was even higher than it was for penetrating vertical imperfections. Error rate for the 1/2-in. transducer remained at zero.

On thinner walled pipe (2 3/8-in. OD by 0.190-in. wall thickness), both transducers effectively differentiated each notch. Fig. 4 shows the details. However, screen readings with the 1/2-in. transducer were very close to the edge of the oscilloscope (near field), which can be difficult to read. We suggest caution when using the 1/2-in. transducer on thinner-walled pipe.

TOO TIGHTLY FOCUSED

Micro analyses of rejections from compression wave wall thickness measurements revealed yet another problem. The transducers can be so tightly focused, a spheroidal inclusion of less than 0.001-in. diameter will be recorded as a rejectable imperfection.

We suggest compression beam transducers be selected as carefully as the shear-wave transmission transducers used to prove-up surface defects.

KEEP IT CLEAN

In examining reject pipe, we found technicians seldom clean pipe surfaces before prove-up.

Steel pipe for the oil and gas industry is either varnish coated or is coated with rust. Varnish or rust can dramatically affect the accuracy of prove-up readings.

Standard practice should be to remove surface contaminants, even dried couplant from a previous prove-up, before prove-up.

A contact couplant should be used between the transducer assembly and pipe surface. The couplant should not become aerated when the wedge is moved along the pipe surface and it should not experience viscosity changes due to temperature variations.

A COMMON MISCONCEPTION

Because they are on the same page, the illustrations for calibration standards in many specifications can lead to confusion.

Despite what the illustrations may lead some to believe, there is absolutely no relationship between an N10 or N5 notch and an 1/8-in. or 1/16-in. drilled hole. In tests, we calibrated 1/4-in. and 1/2-in. transducers at 100% screen height on a 1/8-in. hole drilled through a 16-in. OD by 0.495-in. thick pipe ring. The ring had ten notches ranging from 0.049-in. to 0.006-in. deep.

With the 1/4-in. transducer, a 100% screen height signal response was received from notches as shallow as 0.028-in. And the 1/2-in. transducer produced 100% responses from notches as shallow as 0.034-in.

In similar tests with a 1/16-in. hole, the 1/4-in. transducer produced a 100% reading in a notch that was only 3.2% of wall thickness while the 1/2-in. transducer produced a 100% signal at 5% of wall thickness. This inconsistency dictates the need for additional testing with the 1/2-in. transducer and 1/16-in. drilled hole on different pipe sizes and thicknesses to determine repeatability.

NEW API SPECS WILL HELP

The API 5CT specification for casing and tubing scheduled to be published in 1994 will include a requirement for verification of the accuracy of ultrasonic prove-up procedures and equipment. They must show the ability to differentiate imperfection depths greater than and less than the nominal notch depth. But API does not specify the amount of variation.

In our tests, we found that approximately 0.005-in. will illustrate whether a 1/4-in. or 1/2-in. transducer will differentiate the depths. (Figs. 5 and 6).

RECOMMENDATIONS

Exacting specifications are followed during the manufacture of steel pipe. Each of the specifications was written so the pipe would meet demanding service requirements. Testing is essential to ensure those specifications have been met.

Unless testing equipment and procedures, however, also meet exacting specifications, the results are unreliable. Our tests show the lack of precise prove-up specifications produces inaccurate results. Prime pipe is rejected.

Based on the results we've seen, we also suspect pipe that should be rejected is accepted by inaccurate testing procedures.

To improve the accuracy of ultrasonic prove-ups, we recommend:

A 1/2-in. transducer should be used except when pipe wall thickness is 0.300-in. or less. Then a 1/4-in. transducer usually can produce reliable results.
The wedges used to support the transducer should be machine contoured and given wear studs to ensure the accuracy of the refraction angle and to minimize chances in the angle during prove-up operations.
The calibration notch depth specification tolerance should be greatly reduced and vertical alignment be defined to make calibration and testing meaningful. Test results would be even more reliable if calibration rings with pre-cut notches, such as the ones we use in our plant, were used in the field. The rings would eliminate the multiple inaccuracies inherent in hand-cutting calibration notches in the field.
The prove-up results should not be accepted unless the pipe was cleaned just before the prove-up. It is impossible to obtain accurate prove-up readings from rust-coated or varnish-coated pipe.

Every technician should have materials available to clean the pipe before prove-up.

These changes in prove-up specifications and procedures will produce more accurate, more reliable results. Pipe that is fully acceptable under the specifications will not be rejected. And pipe that should be rejected will be.