Paper01
ELECTROMAGNETIC TEST METHODS ON WELDED
CARBON STEEL TUBING - CAPABILITIES AND LIMITATIONS.
InspecTech®
A Division of InspecTech Analygas Group Inc.
450 Midwest Road, Toronto, Ontario, Canada, M1P 3A9
Tel: 416-757-1179      Fax: 416-757-8096
Email: group@inspectech.ca    Web: www.inspectech.ca

By: A.C. Richardson, InspecTech, Canada

       Murray Rose, Alpha Tube Corp., U.S.A.

       Rick Northrup, Alpha Tube Corp., U.S.A.

Introduction:

Weld seams in carbon steel tubing are routinely tested directly on the tube mill as the product is made; and such testing has become a vital part of any well rounded Q.A. programme for the tube producer. Ultrasonic or Electromagnetic Non-Destructive Testing (NDT) techniques are acceptable methods under most codes and practices.


Ultrasonic testing is the method of choice, but for thin walled or small diameter tubing (less than 3mm or 0.125" wall, or less than 50mm or 2" dia.), ultrasonics are not easily applied to on-line testing. This leaves Electromagnetic methods as the most viable option for testing smaller tubes; and specifically, Eddy Current testing has been unchallenged in the tube industry for several decades.


There is, however, an electromagnetic method of inspecting ferrous products other than Eddy Current, and that is the Flux Leakage (or Diverted Flux) method. Two years ago my colleague, Zbigniew Kaminski, outlined the work we were doing to develop a viable flux leakage test for small tubing as a possible improvement upon the more traditional Eddy Current test. This work has progressed to the point of several working installations, and we are now able to report on some field experience.


Electromagnetic Testing in the Tube Industry:


The standard reference for the NDT industry is " The Non-Destructive Testing Handbook" which now runs to 10 volumes and covers all aspects of NDT. Volume 4, which deals with Electromagnetic Testing, discusses the factors PD, the probability of finding a real defect, and, given that the test system is used at high enough sensitivity to get a useable PD, there must also be PS, the probability of detecting a spurious signal.


Clearly, the tube producer wants to use NDT systems which boast that PD = 1; and PS = 0 for their mills. Sadly, the NDT industry has to date not been able to deliver. Furthermore, the real working relationship between PD and PS is very likely to be influenced by external issues such as operator skill level and mill environment.


With this in mind, the selection of an optimum test method for a given environment can become a complex issue, involving not only how effectively a given system will find defects; but also involving considerations of how well the system will be understood and used in the real world of the tube producer's plant.


Late in 1999 we had the opportunity to operate both Flux Leakage and Eddy Current test systems at the same tube mill, simultaneously inspecting the same product. The results of this comparative evaluation are discussed in the following sections.


Principles of Operation:


Eddy Current (E/C), and Flux Leakage (F/L) are both electromagnetic test methods. In order to examine their relative merits, we need first to look briefly at the principles of operation of both systems.


An Eddy Current test can be performed on any electrically conductive medium. A coil carrying a high frequency alternating current will create electromagnetic fields in a conductor which is close by, and so called eddy currents will flow in the conductor such as to oppose the primary field. If the conductive medium is passing under the probe, and it is uniform in nature, then a stable situation exists. However, if the conductive medium changes in its uniformity (encounters a defect) the electrical equilibrium is disturbed, and the result can be amplified and presented in many different formats.


Fig 1 shows in diagrammatic form the principles of the E/C test.


It is important to note, however, that carbon steel is a ferromagnetic material, and a direct eddy current test on a magnetic tube weld will be almost useless due to meaningless signals from the weld surface. It is therefore necessary to apply a strong magnetic field to the steel to render it the same as a non-magnetic electrical conductor such as copper or aluminum. Hence, the complete eddy current system for carbon steel tubing in fact resembles the cartoon in fig 2, which shows the addition of the magnetic saturator.

It is clear that the eddy current effect is closely related to the proximity of the probe to the test piece, and any variation in that spacing must be minimized. Also the eddy current field decays within the material, resulting in reduced sensitivity through the thickness of the test piece. The relationship is defined by the well known equation:-


1 Where: δ = Depth of penetration (37%)

δ= ---- μr = Relative permeability

_πfσμ σ = Conductivity

f = Test frequency


Note, that the relative permeability is 1 for non-magnetic materials, but over 3000 for unmagnetized steel. This illustrates the need for the magnetic saturator in order to measure anything other than the surface noise in the weld.

Consider now the merits of the F/L test. Instead of having to eliminate the nuisance value of the magnetic permeability of steel, the F/L method actually uses this property in the detection of defects. The technique is as old as NDT itself, but we have only recently started to apply it to weld seam testing in small tubes.

The principle is very simple, and the diagram in fig. 3 illustrates the essential points. If the material is defect free and homogeneous, magnetic flux is distributed uniformly throughout the weld zone. Singularities will distort the uniform field, and cause some stray flux to "leak" at the surface, where it is detected, amplified, and can be presented in a number of different ways.

There is no simple relationship between the effect of depth of defect and amplitude of signal as there is in eddy current testing, although finite element analysis has produced some very good guidelines. However it is evident that the F/L system enjoys a good response to id singularities, with a proportional signal decay with depth. Add to this the excellent "absolute" capability of F/L testing and its ease of calibration and operation, and there are sound reasons for pursuing the development.


A further compelling case can be made for F/L testing where the parent material is galvanized or aluminized, and the weld zone is remetalized immediately after welding and prior to testing. In this situation, there is a thin coating of Zinc or Aluminium for which conductivity on the IACS scale is around 60% covering the steel which is to be tested, which in turn has a conductivity of around 10% IACS. It is clear that the eddy current activity is going to be concentrated in the coating, and testing the steel below becomes a "second layer" problem. The F/L test, on the other hand, responds only to magnetic materials, and treats Aluminium or Zinc coatings the same as air.


Details of Trials:


During the first 2 weeks of December 1999, an Eddy Current test system was installed alongside an existing flux leakage unit on a tube mill which was producing 3.50" and 4.00" carbon steel tubing, with wall thickness in the range of 0.056" to 0.100". The units were both calibrated and supervised by the same factory technician using the same calibration pieces; and the data loggers, which are part of the NDT systems, were used to record total production and defect indications. The data loggers were run only during the time that the factory technician was present and supervising both equipments. During the course of the test, the data loggers totaled 104,700 feet of material tested in this way.


During the trials, both carbon steel and aluminized product were run, but separate records were not kept for each type of product.


Results of Trials:


All defect indications were marked automatically by the separate spray paint systems on the test units. Coupons were cut containing the sprayed areas and these were removed to the Q.A. facility for examination. Defect clusters, occurring at the beginning and ends of coils were ignored in this study. These regions are routinely cropped out of finished product.


In most cases, the reason for the indication was visually obvious. When there were no visual clues, the coupon was crush tested to check for internal defects.


In order to evaluate the results in a manner which would relate directly to tube industry Q.A. practices, weightings on a scale of 1 to 4 were assigned to the indications.


1 for very serious defects


2 for less serious but definitely rejectable defects


3 for marginally rejectable defects.


4 for totally spurious indications.


To be more descriptive, category 1 represented those defects which were visually obvious, or which split immediately on crush testing. Weld misalignment or overlaps being the primary cause.


Category 2 defects could either be seen, or would cause at least a partial failure on crush testing. Small cold spots or mismatches were mostly to blame for these indications.


Category 3 defects may or may not be cause for rejection depending upon criteria used, and included flash effects and heat variations giving a wavy appearance to the HAZ. Category 3's passed crush testing.


Category 4 were indications for which no reason could be found, or were known to be spurious such as inclusions or drops on the aluminized surface.


Table 1: Results of Comparative NDT techniques:



Type Of Indication


1

2

3

4

Eddy Current

3

0

9

24

Flux Leakage

3

10

11

12

Total Indications By Both Techniques

4

10

13

36

Looking back to the probability functions discussed earlier, we can do some calculations on probabilities, which of course are very primitive owing to the small size of the sample and the lack of minute investigation for additional defects which might have been missed by both systems.

If all indications 1 - 3 are considered rejectable and all category 4 indications are considered spurious then there were a total of 27 rejectable indications and 36 spurious indications.


Therefore for E/C testing PD = 12/27 = .44 PS = 24/36 = .66

and for F/L testing PD = 24/27 = .88 PS = 12/36 = .33


Conclusions:


The foregoing analysis of PD and PS is superficial at best, and an accurate determination of these factors in an everyday tube mill situation would be close to impossible. Nonetheless, the margins shown in the results seem to demonstrate quite clearly that the Flux Leakage test method has much to offer.

Other benefits of the Flux Leakage testing method include easier set-up and operator training. As mentioned earlier, these benefits are likely in themselves to reflect favorably an overall performance in terms of PD and PS.


APPENDIX `A'


Analysis of category 2 defects, located by Flux Leakage but not Eddy Current.


Description

Defect #

4" long cold weld, opened at crush

1

Cold weld, opened at crush

2

Cold weld, opened at crush

3

I.D. lack of fusion

4

Pin hole

5

Edge misalignment

6

Edge misalignment

7

Edge misalignment

8

Edge misalignment

9

Edge misalignment

10