 Table of Contents ECNDT '98 Session: Aerospace | First Routine Aircraft NDT with SQUID GradiometerYuri Tavrin, Johann H. Hinken*
|
| TABLE OF CONTENTS |
Two methods for non-destructive testing of aero turbine disks are described. The first method is presently used for routine tests to detect minute ferrous inclusions. The method is based on the detection of magnetic fields with stem from the magnetic remanence of the inclusions.
The second method allows to detect segregations, even if non-ferromagnetic. This is shown as a preliminary result of a feasibility study. This method makes use of spatial variations of the magnetic susceptibility of the disk material. This method has to be developed further on before routine tests can be performed.
The availability of the advanced HMT Magnetometer System allows for two new methods in NDT. Both allow for volume inspection in certain situations where standard techniques like ultrasonic testing fail.
With the availability of such a SQUID system the question arises what it brings to non-destructive testing. One has to check if existing methods can be improved by using this systems. And, one has to check if this system allows for powerful new NDT methods.
Two such new and possibly powerful methods are described in this paper. They allow for the detection of 1. extremely small ferromagnetic inclusions in the volume of a non ferro- magnetic workpiece and 2. non ferromagnetic segregations in the volume of a non ferromagnetic workpiece.
Before describing these methods, in the next paragraph we describe the measuring system.
 Fig 1: HMT Magnetometer system |
 Fig 2: Principle view of second-order gradiometer |
The output of the system is not the magnetic field itself but its second order gradient, see fig. 2. Working similar as an electronic bridge circuit, this second order gradiometer essentially suppresses environmental noise because the output signal is proportional to 1/R5, where R is the distance of the gradiometer unit to the magnetic field source. The device under test (DUT) is usually placed directly below the HMT (R small) and thus a signal from the DUT is practically not suppressed. The noise sources, however, are usually far away enough (R large) to sensure that the magnetic noise is considerably suppressed.
Main system parameters are shown in table 1. The fine field resolution is important. When operating in an unshielded space, however, at least such important is the high common mode rejection and the gradient rejection in order to suppress parasitic magnetic fields.
| Parameter | Value |
| Gradiometer Base-line | 2 x 60 mm |
| Field resolution | 130 fT . Hz-1/2 |
| Dynamic range | 150 dB |
| Slew rate | 105  o / sec
|
| Field/flux coefficient | 5 . 10-10 T / o
|
| Common-mode rejection | 4.000 |
| Gradient rejection | 200 |
In these tests it is to be determined if iron inclusions with a mass greater than a minimum value are present and, if yes, at what location within the disk. The large thickness of the disk is a special challenge. While eddy current and ultrasonic test generally fail we operate as follows.
 Fig 3: SQUID - Signal of ferrous inclusions with different masses |
Fig 4: Signal of remanent magnetization of small iron particle. Mass is approximately 10µg Lift off is 4mm |
Fig 5: Depth and mass determination of inclusion a)Determination of the depth from the signal ratio at different lift offs. Difference in lift off equals to 20mm. b) Determination of the mass of the inclusion from depth and signal amplitude |
A measurement procedure has been developed that allows to determine the mass of the inclusions as well as their locations with respect to radius, angle, and depth (2). For the depth determination use is made of the approximate 1/R3 dependence of the magnetic field strength from the distance R to the inclusion: When in a first measurement at a small lift off an inclusion is detected, the measurement is repeated at an increased lift off. From the signal ratio the depth can be calculated or seen from a diagram like fig. 5a which was generated experimentally. After that, from calibration curves like fig. 5b the absolute value of the signal leads to the mass of the inclusion.
F.I.T. Messtechnik GmbH is performing routine tests of this type on turbine disks. The aero engine manufacturer BMW Rolls Royce is using this technology to asses the cleanliness of certain materials. Based on this NDT technology high performance materials with improved reliability can be used for optimum design.
The non-destructive testing of turbine disks for sub-surface segregations is still an open problem. Eddy current testing does not allow to test deep enough below the surface. And ultrasonic testing often failes because mostly the ultrasonic contrast is not high enough to be detected. Being aware of this problem, the Contaminated Billet Study of the U. S. Engine Titanium Consortium has been performed since 1995 (3). The problem of non-destructive detection, localization and sizing of a sub-surface segregation itself and not of a crack originating from the segregation, however, still seems to be open.
The basic idea for using SQUIDs to solve this problem is the following: Generally, the alloys used for turbine disks are non ferromagnetic. Therefore they are practically transparent for DC and near DC magnetic fields. This gives a chance for sub-surface inspection. These alloys, however, are such weak magnetic that they generally are called non-magnetic. But when looking very sensitively to metals and other materials which generally are called non-magnetic they turn out to be paramagnetic or diamagnetic. Their relative permeability µr is not exactly but very near to one. It is better expressed by the susceptibility 
, with the relation
µr = 1 + . Table 2 give susceptibility values of some elements.
| Element | 104
|
| paramagnetic: | |
| Mn | 10 |
| Cr | 3,3 |
| Pt | 2,5 |
| Al | 0,21 |
| diamagnetic: | |
| Cu | - 0,10 |
| Zn | - 0,12 |
| Pb | - 0,15 |
| Ag | - 0,19 |
| Au | - 0,39 |
| Bi | - 1,8 |
of some elements at
O oC,
Fig 6: Sketch of sample from IN718 containing segregation, placed in measurement system. Fig 7: IN718 sample with segregation, SQUID-Signal versus rotation angle, scan radius is constant and equal to radial coordinate of segregation. Measurements are performed after demagnetization.
|  a) Sequence of about 3 cycles measured from the curved side of the sample.  b) Detailed view of Fig. 7a but with averaging over 4 cycles.  c) As in Fig7b but measured from the flat side of the sample with changed direction of rotation. |
The measurements are performed on a sample made from Inconell IN 718 placed at our disposal by Motoren- und Turbinen-Union Munchen GmbH (MTU). The sample is of disk shape and has a curved side and a flat side, see also fig. 6. On the curved side MTU had detected a segregation reaching to the surface but with unknown depth. It had been detected by segregation etching and its existence was confirmed by eddy current testing. Both methods are successful only in such a special case where the segregation extends to the surface.
We demagnetized the sample, placed it on the turntable of the HMT Magnetometer System and scanned it at a radius which is equal to the radial component of the segregation. Fig. 7a shows the measuring result over about three cycles. The homogeneous external field, which is distorted by the disk with its matrix material and its segregarion, in this case is the earth magnetic field. Fig. 7a predominantely shows a periodic signal with the period of 360o. This stems from the improperly adjusted turntable. Furthermore the rotation of the turntable was not smooth resulting in a superimposed nearly random signal. The signal, however, which is marked with an arrow is not random and occured directly above the segregation.
Fig. 7b shows the detailed view of fig. 7a but with averaging over four cycles. The random signals have largely disappeared, the signal over the segregation remains and is more clearly to be seen. At this point of the examination, the signal can be caused by a susceptibility variation or by a ferromagnetic inclusion which is not fully demagnetized. In order to distinguish between them, the disk was also measured from the flat side with the direction of rotation being changed in order to have the measuring plots better comparable. Fig. 7c shows the result. The sign of the signal, which is marked with the arrow is the same. If the origin would be ferromagnetic remanence, the sign would have changed. Comparing fig. 7b and c shows that the origin of the signal is a susceptibility variation, either paramagnetic or diamagnetic.
Most important is the following: Fig. 7c shows that the segregation can be detected also from that side where it does not extent to the surface. This confirms that the material is transparent for this kind of susceptibility measurements. That means, that also segregations can be detected, which are in the volume and do not extent to the surface.
Before applying this method to routine testing of critical parts like turbine disks, it has to be further developed and its limitations have to be checked carefully.
The first application has been fully developed and is performed in a routine service: Turbine disks are tested for extremely small iron inclusions. The physical principal is to measure their magnetic remanence.
The second application was checked in a feasibility study. The preliminary results show that segregations in Inconnel IN 718 can be detected within the volume. It can be assumed that segregations also in other type of material can be detected by this method. Several details, however, have to be checked before a routine use of this susceptibility based method. Among these details are to determine the susceptibility contrast of various segregations in various matrix materials as well as sizing and localisation procedures of the segrations.
| NDT.net |