Eddy Current Testing
Coming of age
By Jim Cox
During the period from 1775 to about 1900, several scientific experimenters investigated the many facets of electromagnetism. If we think about the tools that they must have used in that time frame we will soon realize that this stuff called eddy current testing must be pretty simple. They didn't have computers, electric lights, or even 110 power to plug into. A couple of guys sitting around with some magnets, a compass, a crude battery, and some different pieces of metal were able to figure out what it is and come up with the rules and formulas that we still use today!
Figure 1 shows a model of a magnetic field starting to move across a block of electrically-conductive material. You can do the same experiment with a magnet on a string. This was first documented by a man named Arago. He discovered that the motion of a swinging magnet was rapidly slowed when a nonmagnetic, electrically-conductive block was placed under the magnet. (Hint: try both stainless steel and copper.) This is an eddy current (EC) test. The relative field strength and the number of flux lines cutting into the metal depend upon the position of the magnet.
Over a given time period there is a changing magnetic field at any given point in the metal block. The changing magnetic field strength forces electrons in the material to start moving, i.e., current flow. When a current is flowing in a conductor it will also create a magnetic field around itself (Lenz's Law -— the Right Hand Rule). The (eddy) current flow in the block produces a "secondary" magnetic field that is opposite to the field that created it. These oppositely charged fields (north vs. south) will attract each other. The magnet swing is slowed based on the attractive force between these two fields. This simple model is the basis for some automobile speedometers.
If any of the test or material properties (such as conductivity, permeability, or geometry) are changed, then the amount of eddy current flow created in a metal specimen must also change. If the strength of the eddy current flow changes then the strength of the secondary magnetic field created by the current flow must also change. Since these two magnetic fields are oppositely charged, variations in the secondary field strength cause a shift in flux distribution and strength of the primary magnetic field around the coil. This, in turn, changes the coil's electrical parameters. It is a cascade event. As one part of the picture changes, everything else shifts in response.
One of the historical arguments against using eddy current testing is that it is too sensitive. The signal that we detect during our inspection might be created by some combination of material changes. The possible variables include the material's conductivity, or its magnetic permeability, and/or the geometrical factors of how we are performing the examination. How do we figure out what really changed? Conductivity (s) is an electrical property. It determines how well electrons will move through a material. Metals are normally classified as conductors. Theoretically, we could do an eddy current test on any piece of metal. Resistivity (r) is another term that we can use to describe the electrical properties of a material. Resistivity is more common in the formulas shown here.
Permeability (m) on the other hand is a magnetic property. Not all metals have significant levels of permeability. The permeability value of a metal suggests how it will alter a magnetic field moving through it. Metals with high permeability values (carbon steels) restrict the inspectable area to the upper skin layer with a given probe. Fortunately, the largest percentage of aircraft structures and surfaces are made from nonferromagnetic materials. Their relative permeability is assumed to be one. This means that as the primary magnetic field moves through these materials their inherent permeability does not alter the magnetic field energy level.
The third major test variable, geometry, depends on the type of probe being used. Standard terminology for surface scanning applications include lift-off and edge effect. For either bobbin or encircling coil probe styles the terminology addresses fill factor and end effect. Geometry responses are considered to be a limiting factor in our ability to detect true material changes. This type of response could lead to a poor signal-to-noise (S/N) ratio. Anything that we can do to limit geometry responses from our field of view will improve the quality of the test.
Primary / Secondary Force Relationships
The coil becomes the tool that determines how well we can link our coil's energy into a test piece. The choice of coils becomes critical to how well we can perform a test. The magnetic field produced about a coil is directly proportional to the applied current magnitude (amperage), the rate of current change (frequency), as well as the coil's physical parameters. Coil parameters include factors such as inductance, diameter, length, thickness, number of turns of wire, and core material.
With a primary alternating (AC) current (Ip) flowing through the coil, a primary electromagnetic field (fp) is produced about the coil. By placing this excited test coil on (or near) an electrically-conductive test object we will begin to generate eddy currents (IE) in the test object.
A term can be developed to describe the distribution of current density in the test object. Consider the eddy current generated at the surface (right under the test coil) to be 100 percent of the available current. The point in the test object thickness where this current is diminished to 37 percent of its original value is known as "one standard depth of penetration" (d, delta). Figure 2 (on previous page) shows how the available energy is rapidly decreasing as we move deeper and deeper into the test sample. As the energy level decreases we are less likely to detect a change in the material at that point.
Frequency Choice Frequency is going to be the main tool that we use to define the "field of view" during our inspection. The practical field of view is considered to be about 2.5 to 3d (three times the calculated depth of penetration). If our test is designed to look for small stress cracks on the upper surface, then our decision is probably going to be for a smaller coil diameter with a fairly high frequency range. We would want to keep our energy in the area of interest to maintain good flaw sensitivity. If we know that we need to put more energy deeper into a material to detect back surface or second layer conditions, then we are going to have to select a lower inspection frequency. How do we decide what is a reasonable frequency choice for this type of test?
A simple formula using the material's electrical properties (p) and its thickness (t), where "f" is in hertz, "t" is in inches, and "r" is in micro-ohm centimeters (mWcm):
f = 4(p) / t2
Because of the way the formula has been arranged we can disregard the issue of mWcm being in the metric system. We simply look up the resistivity value on that scale and plug in the number. I normally round off the resistivity for all aluminum to about 5. If we assumed that we wanted to detect a back wall crack on a piece of aluminum 0.2-inch thick we would get:
f = 4(5) (.2)2 = 20 (.04) = 500 Hertz
If we plug that frequency into our depth of penetration calculation, you can see that all we have done is put the position of d at the bottom of this plate. That means that we would still have 37 percent of our original surface energy when the primary magnetic field creates eddy current at the back surface. That should give us plenty of detection capability.
Test coils can be categorized into three main mechanical groups. These are probe coils, bobbin coils, and encircling coils. Each of these three basic designs might be driven or operated in one of three modes: absolute, differential, or hybrid. The coil configuration and mode of operation would tell us what type of screen response to expect.
Most of the coils that we use in aircraft component inspection would be addressed as "probe" coils. We are normally scanning across the surface of a material. The mode of operation for surface crack detection would typically be either absolute or differential, based on how the coil or coils were connected to the tester. A differential coil pair is normally used to reduce the effect on non-flaw related responses (such as edge effect).
For most subsurface flaw detection we find that a hybrid or driver-pickup coil tends to provide a better response. The hybrid coils are usually designed to provide a higher S/N ratio in a specific application. If we were trying to get better detection or characterization from second- or third-layer corrosion, a hybrid coil designed specifically for that type of test would probably perform better than selecting a normal surface probe and trying to do the same test.
Weld inspection with normal surface probes would be extremely difficult. Responses from the weld heat-affected zone (HAZ), localized conductivity changes due to recrystallization, weld metal content vs. base material composition (filler wire), variations in permeability, and variations in geometry (weld splatter, surface condition of crown, etc.) would all be occurring simultaneously. Probes have been designed to compensate for these variables that still maintain their sensitivity to cracks.
Eddy current coils are usually optimized for certain frequency ranges. This is due, in part, to certain physical design requirements of probes. In order to obtain an adequate flux density, for example, a low frequency probe requires a larger amount of wire compared to a high frequency probe. Very often, the designing of a probe becomes a compromise between several different factors, i.e., operating frequency, size limitations, flaw sensitivity, and durability.
You have already won half the battle if you can calculate a reasonable test frequency and choose a probe style that is going to provide you a reliable flaw response in the area of interest.
Eddy Current Testing
Coming of age
By Jim Cox
continued form page 1....
Signal Displays/ Calibration
Electromagnetic variations are detected by the eddy current system and displayed on some type of read-out mechanism. Meter-based testers are the simplest to use but have definite application-specific limitations. Cracking and corrosion issues for aging aircraft sometimes require the use of impedance plane (X-Y) displays.
When we are performing an eddy current test on an X-Y system we watch a dot on a screen. As long as that dot doesn't move that tells us that nothing is changing in the test environment. If the dot does move, we now have to try to understand how much the dot moved (signal amplitude) and in what direction the dot moved (its phase). If we can do that, then we can probably estimate "why" the dot moved. Hopefully, we can quantify that electromagnetic change to a specific change in the material properties. With that information, we can make our decision about whether or not this part is acceptable or rejectable.
In order to do that we have to go through a "calibration" process when we start this test. We need to be able to recognize certain types of signals and organize them on our read-out mechanism. This will create a logical, repeatable pattern of responses so we can train our visual and mental process to recognize the differences between signals that are "flaw-like" and those that are normal or expected non-flaw information for an acceptable specimen.
We are always going to have some geometry response as we move the probe across the test specimen. We always try to limit our visual sensitivity to geometry by putting that response at a known position on the screen, normally horizontal. Permeability changes are not a major factor in most aircraft examinations. That only leaves conductivity as a variable. Cracks and corrosion exhibit a change along what is referred to as the conductivity line. Once we understand this relationship, it is a fairly straightforward process to decide what is a true flaw and what isn't.
Any discontinuity that appreciably changes the normal eddy current flow can be detected. Discontinuities (cracks or corrosion) generally cause the effective conductivity of the test object to be reduced. If we look at the arrangement of signals on the screen in Figure 3, this would mean that any flaw would be represented by a positive or upward response on the screen. Any response that was only side-to-side (horizontal) would be related to a liftoff or geometry-related event. Discontinuities open to the near-surface would be much more easily detected than subsurface or far-surface discontinuities.
The displayed signal change due to discontinuities will vary based on the size (volume) and location (phase) of the signal source. Phase is normally equated to depth or location within the specimen. Voltage is more dependent upon total volume (length, depth, width) of the discontinuity.
In the test results shown in Figure 4 we would need to look at our procedure. We may not be meeting the required flaw sensitivity. The tester has been "calibrated" properly, but it is likely that the frequency choice was not correct. If we need to be able to detect cracks throughout this plate, then the fact that we cannot get a reasonable response from crack C is a major problem. Notice that at whatever frequency we were operating, we had a depth of penetration (1d) of about one-fourth of the way through the block. When we scan over crack C we have very little energy left (3d= 5 percent). We used 95 percent of our energy just to reach this point in the block. What if we were to decrease our operating frequency? Wouldn't that give us more depth of penetration into the sample? This should put more energy deeper in the plate and we should get a larger amplitude response from crack C.
That would not be the only signal change. Remember that since we are working at a new depth of penetration, the phase angle of all of the responses would be different. The advantage would be that the new phase angle response of crack C should be more like the original response that we saw from the B indication at the original frequency. By running at a lower frequency we have dramatically improved our test for crack indications throughout this plate.
Let's wrap it up
What can we really say about eddy current testing?
• Eddy current testing is not a black-box art form. It has many different applications. It can provide material sorting capability, it can measure material properties (i.e., conductivity in percent IACS), or it can be used to detect different types of service-related damage in a wide range of structures.
• ECT can be used on many different metallic surfaces. Within reason, you don't need to remove paint or other protective coatings to perform the test.
• ECT works best for near-surface indications but can also be used to detect subsurface cracking and corrosion. We don't normally concern ourselves with sizing front surface flaws. Once you detect a surface crack, the acceptance criterion becomes one of location and probability of crack length extension. It is possible to set up sizing criteria for subsurface corrosion issues. This is becoming more and more common in aging aircraft inspection procedures.
• ECT requires trained personnel to understand it and apply the techniques properly. This is especially true if you don't have a dedicated procedure defining exactly how to set up a given test system for a given application. A good manufacturer's applications lab can assist you in making the right choices or at least let you know if the test is feasible. Some provide this service free of charge.
There are a lot of misconceptions about the limitations of eddy current testing. With the level of equipment and probe technology that has been achieved in the last few years, it is now possible to apply ECT to many test scenarios that previously could not be handled. Weld inspection for near surface flaws is a good example of this situation. With the right probes, both ferritic and non-ferritic weldments can be inspected. Another advancement that has taken place is the application of filtering and mixing technology to suppress unwanted non-flaw information.
Many aircraft inspection codes do not give eddy current its rightful place in the market. I suspect that this is normally based on the fact that no one wants to go back and re-address all of the paperwork that is required to upgrade those documents. When an EC test is applied with the proper probe and frequency choices, it can easily surpass the detection capability of many other NDE inspection processes. When the code phrasing uses something like ". . . demonstration of detectability equal to . . ." then you have the option of applying the most cost effective processes. Eddy current testing should come to mind at that point.
I have seen procedures that call for the same penetrant (PT) test to be performed multiple times to attempt to find tight cracks. The ECT inspection time would be much less and the flaw detection potential would be much higher in many of those cases. The same could be said for some inspections requiring magnetic particle (MT), or radiography (RT). Because a particular NDT method is listed as the code inspection does not necessarily mean that it is the best detection technique. Each situation should be evaluated on a cost vs. detection probability basis.
I doubt if Dr. Arago had any idea where his swinging magnet would take us. If you are not using ECT in your inspection processes on a regular basis, it may be time to get reacquainted with a very old technique that has truly come of age.