The alternating current potential drop method for surface crack depth measurement

TWI Bulletin, January 1982

by A C Duncumb, BSc and P J Mudge, BSc

Mr Mudge is a Senior Research Metallurgist and Miss Duncumb a Research Physicist in the Non-Destructive Testing Research Section of the Engineering Department.

Surface breaking cracks are one of the most detrimental defects which can exist in a structure. When they are found, the structural significance of such defects can be assessed only if it is possible to measure their depth accurately. The AC potential drop (ACPD) method was considered to be a promising means of performing this task and recently The Welding Institute conducted a programme of experimental work to assess both the accuracy achieved by, and factors influencing, ACPD crack depth measurements. The results described in this article suggest that there is an application for this technique for site measurements of surface breaking cracks.

Potential drop techniques for measurement of the depth of surface breaking cracks and also for monitoring crack growth in fatigue specimens have found a wide acceptance^[1]. The DC method has been more widely used, particularly in the laboratory environment, where much testing is concerned with small specimens. It involves passing a direct current through the volume to be examined and measuring the potential drop across it. A flaw gives rise to an increase in resistance because of the reduction in cross sectional area. This is superimposed on the potential gradient between the current input and output points, and potential differences can be calibrated in terms of defect depth. The size of potential drop depends on the defect area in relation to the current density in the specimen across the defect plane. Adequate current densities can be achieved in small samples, but very large currents are required to produce sufficiently large current densities for flaw detection in larger sections. The method is therefore impractical for large structures.

An alternative approach is to use the AC method where, as a result of the 'skin effect', the current flow is restricted to the surface of the specimen under test. Here, potential drop because a defect is present does not result from a reduction in cross sectional area, but from an increase in current path length around the flaw. Thus a much smaller alternating current is required for a given voltage drop across the specimen surface than with the DC technique, rendering any apparatus adopting this principle more readily applicable to a wide range of components.

The study described below was designed to examine the influence of test parameters on the performance of the ACPD technique and to assess its accuracy for measurement of the depth of surface breaking cracks, using a commercially available equipment - the Unit Inspection Crack Microgauge. The experimental work was divided into two stages. In the first, a plain plate specimen with a tapered machined notch in the upper surface was used to establish optimum conditions for testing, and the effect of different operators using the equipment. The second stage consisted of applying ACPD tests under the derived optimum conditions to specimens of geometries typical of those found in structural steelwork in which fatigue cracks were grown. This was aimed to determine the accuracy of the technique for measurement of such cracks at welds.

Principles of operation

To perform a potential drop measurement, a steady state electrical field has to be established in the test area and a probe is required to measure potential differences at points within the field. An alternating field is set up around a defect by passage of AC between two input points, and the skin effect restricts most of the current to a thin layer of the metal surface. For a given alternating current of frequency, f, 63% of the current flows in a skin of depth, δ, where:

This skin depth can be altered by varying the frequency. ^[2]

Assuming that the current inputs are widely spaced, so that a current flow essentially perpendicular to the suspected crack is maintained, and that the measurement probe points are placed along a field line, the voltage drop between two closely spaced points in the field is proportional to the current path length between them. As explained above, a surface breaking crack alters this path length (the current effectively travelling around the crack), thereby changing the voltage drop across the crack mouth at the surface. Assuming that the crack depth, d, is an order of magnitude less than its length, then d may be calculated from:

Details of the theoretical derivation of this expression are given in ref.^[3] and ^[4].

Experimental

Equipment and test method

The Unit Inspection Crack Microgauge has simple controls and digital readout of potential drop, see Fig.1a. Current input is via spring loaded pins clamped to the specimen by magnets. To avoid interference effects, the current input leads were kept well apart from the probe lead (recommended in the manufacturer's operating instructions) and the instrument was earthed to the testpiece. Adequate field current is denoted by a green lamp on the instrument front panel. Should the resistance between current inputs be too great to allow a field of sufficient strength to be developed (resistance in excess of 4ohms) no digital readout is obtained and a red panel light warns the operator.

Voltage drop measurements are made by a small two pronged probe. Such probes may be a variety of shapes for different applications but probe tip spacing is kept roughly constant at around 10mm. The Unit Inspection standard and corner probes were available for this work, see Fig.1b. The corner probe is designed for examination at rightangled changes of section due to its angled contact pins and is thus particularly suited to weld testing. The potential drop across the probe pins is amplified and presented on an LED display and is also available as an analogue output suitable for input into a chart recorder.

The only calibration required prior to testing is to check on-site that the readings obtained before and after rotating the probe through 180° have the same absolute value. This equality is attained by adjusting a calibration control until a balance point is reached. Once found, the instrument is correctly set up to perform a test and the balance point requires only periodic checking.

With the current input points equidistant (about 250mm) from the predicted area of the defect, two measurements of the voltage drop are made: (i) between the probe pins on plain material adjacent to the defect; and (ii) with the probe pins straddling the defect. The crack depth is calculated from [2].

Notched plate specimen tests

All specimens for these trials were manufactured from BS 4360 grade 50D steel, a widely used structural grade. The specimen was a 300 x 580 x 25mm piece of plain plate, into which a straight sided notch 0.8mm wide was cut, tapering from 6mm to zero depth across the full 300mm width of the plate. This notch was used for calibration purposes and to derive optimum testing conditions. The notch depth was measured at a number of points to determine the profile under a variety of conditions, the three major variables considered were:

1. surface preparation requirements;
2. optimum current input positions relative to the notch;
3. the effect of using different operators on the test result.

Surface preparation was studied by commencing with the specimen in the as-received condition and improving the surface at the current input and voltage measurement point in stages. The effect of current input position was investigated using various combinations of the current input points, labelled A to M in Fig.2. Both proximity of the current inputs to the notch and asymmetry of the field were examined. The influence of different operators on the outcome of the test was examined by using four technicians having varying degrees of familiarity with the equipment. Each measured the notch depth, using a fixed test procedure and under supervision to ensure correct operation of the equipment.

Results were compared with a mechanical measurement of notch depth using a dial gauge.

Welded specimen tests

Specimen 1 was a plate containing a flat double V butt weld manufactured using the manual metal arc (MMA) process with a 60o included angle preparation. The overall dimensions were 581 x 300 x 25mm, the weld being mid-way along the 581mm length, as shown in Fig.3a. All but the central 100mm of each weld toe was ground to inhibit crack growth away from the specimen centre when it was fatigue cracked. Eleven measurement positions were marked down each side of the unground part of the weld, as shown in Fig.4.

Specimen 2 was a full penetration T butt weld with an attachment MMA welded (double sided weld, leg length 25mm) to a base plate (Fig.3b). Again, all but the central 100mm of the weld toes was ground to promote fatigue cracking in the centre of the specimen. Potential drop measurement positions 10mm apart were marked on the specimen and current input points are shown in Fig.5.

Specimen 3 was a sector of a girth weld in 457mm diameter pipe. The weld was located mid-way along the specimen (Fig.6). Again, the central 100mm section of each weld toe was left unground to promote fatigue cracking in this region. Current input points and potential drop measurement positions were similar to those in specimen 1.

On all three specimens the current input points and potential measurement points adjacent to the weld were lightly abraded to achieve good electrical contact, but the weld surfaces were not cleaned, as to do so would have altered the as-welded profile. Current inputs were placed symmetrically about the weld, 250mm from the centreline.

Specimens 1 and 3 were subjected to fatigue loading, No. 1 in three-point bending and No. 3 in tension. ACPD readings of crack depth were taken prior to application of a fatigue load (as a control measure), at the first visible sign of fatigue cracking, and thereafter at intervals of 20,000 load cycles until failure. On resumption of fatiguing, after an ACPD measurement had been taken, 2000 cycles at low stress were included to mark the crack front position at that time by a 'beach mark' line of differing fracture appearance. This facilitated subsequent comparison of actual and estimated crack depths at the various measurement points.

Problems were experienced with fatigue cracking of specimen 2, the T butt weld, because of its geometry. To establish whether the T geometry had an effect on the ACPD measurements, a fine notch 0.15mm wide and 3.5-4mm deep was milled along the weld toe on one side of the T. This was then measured using a similar ACPD procedure to the other two welded specimens.

Results

Notched plate specimen

Surface preparation was found to be important from the point of view of maintaining good electrical contact for both current and voltage probes. The attachment of the current prods by magnets proved to be simple and convenient, but the contact surface between magnet and workpiece is also required to be free from dust and scale, otherwise poor adhesion occurs. If there is poor contact between the current inputs and workpiece the instrument warns the operator, and if poor contact is made at the voltage measurement probe a stable readout will not be obtained. Both circumstances are fail safe, in that a measurement cannot be made under such adverse conditions.

Another source of poor contact was blunting of the chisel shaped pins on the voltage measurement probes. Presumably, the sharp point penetrates thin films of surface oxide, ensuring good contact, while a more rounded tip, will not. Both the standard and corner probes performed equally well, but the corner probe was easier to use at a weld toe because of its more favourable shape for this geometry.

Figure 7 shows results from ACPD depth measurements on the machined notch obtained by four different operators with current inputs at A and B (Fig.2). The figure shows a general consistency of estimated depth for each measurement position, indicating that the technique is relatively insensitive to operator variability. Furthermore, the curves in Fig.7 are typical of all results obtained from this specimen, suggesting that, for simple geometries at least, the field input points in relation to the defect are not crucial. There is good overall agreement between estimated and actual notch depth, the large vertical scale in Fig.7 exaggerating the measurement error. The trend is, however, to underestimate notch depth, all values obtained being in the range -0.1 to -1.0mm of the actual depth. For tests conducted by a single experienced operator, the range of measurements was +0.5 to -0.6mm of the true value. This compares very favourably with the accuracies of other NDT methods for crack depth measurement, e.g. ultrasonics, which exhibit an average error of ~1mm with a wider scatter of results.^[5]

The results for probe positions 3 and 4 exhibited lower measurement errors than elsewhere on the plate. This was observed over the whole range of tests applied to this specimen and appeared also to be operator independent. The most likely cause is an asymmetrical distribution of the field in the plate, which is not accounted for in the simple conversion formula given in [2]. This effect is probably a result of the geometry of the notch. The asymmetric distribution of the field also seems to be largely independent of the position of the current inputs. Each measurement position tended to produce the same depth estimation irrespective of the test applied.

In the light of these results the current inputs were placed symmetrically on subsequent specimens and at a distance of about 250mm from the suspected crack, as these positions appeared to give a reduced scatter of measurements. This good practice was also in accordance with the manufacturer's recommendations. A symmetrical field is likely to give better results, other things being equal, as this is one of the assumptions from which the simple crack depth calculation is derived.

Welded specimens

Results from ACPD measurements at intervals throughout fatigue crack propagation until failure were obtained from two specimens - No. 1, the flat butt weld (fatigue loaded in bending), and No. 3 the curved butt weld (fatigue loaded in tension).

Estimates of depth at each measurement position were combined to produce a schematic view of crack profile for each depth at which it was measured. This could then be correlated with the actual depth of the crack at the same point in its fatigue life as revealed by the beach marks on the fracture face (Fig.8). Figure 9 shows such a correlation for specimen 1. There is good agreement between actual and estimated crack depth when a small crack, 1-2mm deep, exists, indicating that small defects could be detected and their depths measured relatively accurately. Similarly, the final measurement, i.e. the crack size prior to the subsequent extensions to failure, showed good agreement between measured and actual values.

However, poor correlation was obtained for intermediate values of crack depth, depth being underestimated significantly. This could be caused by limitations of the equipment but it is more likely to be attributable to two other factors. Firstly, precise positioning of the beach marks on the fracture surfaces was difficult, in most cases the marks themselves being difficult to see. Therefore, errors in the actual crack depth measurement could be quite large, of the order of 1-2mm.

Secondly, a phenomenon often associated with fatigue cracks is that of partial crack closure when the fatigue load is removed. This would tend to provide a conducting path across the crack, thereby producing an apparent shortening. In the three-point bend configuration used to test specimen 1 it was not possible to apply a static tensile load to establish whether this occurred, so that the effect of crack closure could not be quantified, but this was done for the specimen fatigue loaded in tension, as described below.

Two sets of results for specimen 3 were obtained by use of both the standard and the corner probes. One of the fracture faces of this specimen is shown in Fig.10, while Fig.11 shows the correlation between actual and estimated crack profile for the two probe types for several defect depths during the fatigue life of this specimen. As for the previous specimen, a similar systematic tendency to underestimate crack depth can be clearly seen. The same sources of error in the measurements apply here, but an attempt was made to evaluate the effect of partial crack closure on depth measurement by performing the ACPD test with a static tensile load on the crack. This caused the measured depth to be increased by up to around 0.5mm, which improved accuracy, but did not account for the entire discrepancy between actual and estimated values. An example is shown in Fig.11.

On the T butt specimen, No. 2, all tests were carried out using one current input on the base plate and one on the attachment. The fillet weld geometry made it necessary to use the corner probe with angled contact tips to obtain good electrical contact. Figure 12 shows results from these tests. It had been noted during earlier trials that operators adopted one of two positions for the across defect reading; one 'higher up' the weld and one 'lower down' (shown schematically in Fig.13). To quantify this effect each measurement was taken with the probe in both these positions, accounting for the two sets of results displayed.

In each case the Microgauge produced good agreement with actual notch profile, depth measurements within 1mm of the actual value being obtained. Variations in depth measurement resulting from positioning of the probe on the weld cap can be explained by changes in current path length caused by placing the contact pins at the two different locations. However, these measurement variations appeared to be small, typically less than 0.5mm (Fig.12).

Discussion

A major practical advantage of the ACPD crack depth measurement technique is that the equipment is simple to operate. It appears that, provided the recommended setting up procedure is employed, the technique (using the equipment studied) is capable of producing depth measurements of surface breaking defects with an accuracy exceeding that of conventional ultrasonic techniques. The apparatus is also relatively small and readily portable, which makes the technique attractive for site use where ease of access and minimum reliance on the subjective interpretation of the operator are of importance.

However, the simplification of the test procedure and minimisation of operator intervention imply a greater reliance on equipment design and performance, and do not eliminate the need for certain assumptions relating to the behaviour of AC electric fields in components of non-uniform geometry. Also, it is not possible to account for the influence of variations in operating practice from user to user. This programme has sought to quantify the performance of the ACPD technique with reference to some of these considerations.

Of prime importance to the accuracy of this technique is a detailed knowledge of the distribution of AC electric fields in components of a given geometry, and the effect which the presence of a defect has on the field. Theoretical work, such as that described in ref.^[3], has enabled formulae to be produced by which potential variations within the field can be directly related to the defect depth, at least for simple geometries. Tests on the notched plate specimen showed little effect of an asymmetrical field and generally the effect of specimen shape for flat, slightly curved and T configurations was insignificant. However, previous experience at The Welding Institute has shown that on more complex or highly curved geometries, such as joints between tubulars, the simple formula used for all results presented here (equation [2]) is no longer valid, in which case more appropriate corrections have to be used.

Again, within the limited scope of these trials, variation in results produced by different operators was acceptably low, indicating that the instrument effectively removes the onus on the operator to make a correct judgement of crack depth based on the data presented to him. However, this does not diminish the importance of good practical techniques.

Surface finish requirements appear to be confined to removal of scale and rust to achieve good electrical contact at current input points and also at positions where the voltage measurement probe is to be applied. As noted above, a measurement could not be taken without adequate electrical conditions being attained. Beyond this, surface finish appears to be immaterial.

Overall accuracy of depth measurement of fatigue cracks being grown in specimens 1 and 3 was not as good as that obtained on the artificial notches. On-site, the information most frequently required is whether a crack exists or not, and its approximate depth. The results in Fig.9 and 11 clearly show that the cracks were detected at an early stage and their extension was monitored in spite of the tendency to undersize. The relatively good accuracy of this method compared with conventional ultrasonics also indicates that it could be employed for monitoring fatigue crack extension in structures. Measured vs actual sizes for the two welded specimens, 1 and 3, are shown in Fig.14. The tendency to overestimate the depth of very small cracks and generally to underestimate larger ones is due in part to the technique predicting an average crack depth when this is changing rapidly, for example at crack ends, (see Fig.9 and 11). However, it should be noted that high degrees of accuracy for crack depth measurement can be achieved if more specialised ultrasonic techniques, such as time of flight measurement are used. ^[6]

On a real structure the effects of crack closure are probably not so apparent, as a live load is normal being continuously applied while the test is carried out.

Conclusions

1. Tests carried out on artificial notches demonstrated that reliable estimates of depth, within +0.5 and -0.6 of the true value, could be achieved by different operators using an ACPD instrument.
   
   
2. For a number of relatively simple geometries the location of the AC field inputs was not critical, and a simple conversion formula could be applied relating potential difference to crack depth.
   
   
3. The instrument concerned removed much operator subjectivity from the test. However, application of the technique does require good practical procedures and a knowledge of the limitations of the method on more complex structuralsections.
   
   
4. Measurement of fatigue cracks was subject to larger errors but these compared favourably with other crack depth measurement techniques.
   
   
5. The initiation and propagation of a crack at a weld toe could be monitored using this method.

Acknowledgement

The authors thank BP International Ltd for permission to publish this work.

References

1. Slater G: 'Fatigue crack growth monitoring - a review of techniques'. Welding Institute Research Bulletin 1981 22 (3) 66-69.
   
   
2. Bleaney B I and Bleaney B: 'Electricity and magnetism'. Oxford University Press, Third Edition, 1976.
   
   
3. Dover W D et al: 'The use of AC field measurements to determine the shape and size of a crack in metal'. Publ Unit Inspection Company.
   
   
4. Dover W D and Collins R: 'Recent advances in the detection and sizing of cracks using alternating current field measurements'. Brit J NDT 1980 22 (6, November) 291-295.
   
   
5. Jessop T J et al: 'Size measurement and characterisation of weld defects by ultrasonic testing - Part 2: Planar defects in ferritic steel'. Welding Institute Report; to be published.
   
   
6. Mudge P J and Whitaker J S: 'Depth measurement of cracks in toughness specimens - use of an ultrasonic time delay system'. Welding Institute Research Bulletin 1979 20 (7) 198-203.