Non-Destructive Test Methods for Hollow-Core

Composite Insulators
Armin Merten
Department of Research & Development
Reinhausen Power Composites GmbH
D-93057 Regensburg, Germany
Email: [email protected]

AbstractSeveral test methods were evaluated in order to between tube and flange will not lead to electrical problems,
prove their suitability to detect voids, air bubbles, and adhesion this paper concentrates on the investigation of the sheds, the
failure in hollow-core composite insulators. Due to the relatively interface between sheath and tube, and the glued scarf joint.
complex geometry of the insulators (bowed surface, shed-profile)
and the criticality of even comparably small defects in high- II. N ON -D ESTRUCTIVE T EST M ETHODS FOR P LASTICS
voltage equipment, most of the methods did not provide satisfying
results. Most promising seem to be Computed Tomography and This sections describes various non-destructive methods
a rather new method, the so-called Micro-Wave Based Detection. to detect irregularities in plastic material. Experiments have
been performed with only four of them. The two others (i.e.
I. I NTRODUCTION
Thermography and Laser-Shearography) will be evaluated in
Inhomogeneities like voids, air bubbles, or impurities should future work.
generally be avoided in electrically insulating materials in
order to prevent partial discharges, electrical breakdown, or A. Ultrasonic Detection
- if applicable - mechanical failure. Although in a well Ultrasonic Detection is a well established method in non-
controlled production process the aforementioned defects do destructive testing not only for plastics but also for all other
not occur, it could be useful to be able to find imperfections kinds of material. The results exhibit a high reproducibility if
without destroying the part to be investigated, i.e. during an the investigator is well experienced. Furthermore, it is possible
homologation process or in order to find the origin in case of to automize this method e.g. for continuous quality control.
failure. The measuring principle relies on the dependence of the
Non-destructive testing methods for plastic components are speed of sound in the medium to be investigated. If an acoustic
widely available for quality assurance and fault diagnostics. wave propagating through a material with acoustic velocity c1
Most of the methods described in the following are approved meets an inhomogeinity with acoustic velocity c2 6= c1 , the
to be suitable particularly for parts with a simple geometry wave will be scattered and partially reflected. Knowing the
and with plain surfaces. However, in case of hollow insulators runtime of the wave through the homogeneous material and
for high-voltage equipment, the situation is much more com- measuring the time of the reflected fraction to come back, the
plicated because of the circular cross-section of the insulator position of the defect can be determined [1].
and the wavelike sheds on the outside. Figure 1 illustrates The costs for a commercially available system are moderate
schematically the setup of a typical hollow composite insula- and defects > 1 mm can easily be detected. However, the
tor. application of this technique to hollow insulators is restricted
for the following reasons:
Due to the sheds, the testing probe can not be attached
to the outer surface of the insulator.
Special probes are necessary to fit the curvature of the
tubes. Therefore, for each diameter of an insulator another
probe is required.
For longer parts the access to the center area inside the
insulator is difficult.
Fig. 1: Schematic view on the composition of an insulator The resolution is insufficient to detect critical defects. In
particular, spots of weak adhesion between the sheds and
There are several critical spots at an insulator which are the tube are almost impossible to find.
vulnerable for mechanical or electrical failure, e.g. the bonding
between the flanges and the tube, the bonding between the B. X-Ray
sheath and the tube, and for long assembled insulators the The consistency, the density, and the thickness of the part
scarf joint between the partial tubes. Since flaws in the bonding investigated determine the degree of absorption of X-Ray

978-1-4244-8286-3/10/$26.00 2010 IEEE 536

radiation. The picture obtained as a result of the X-Ray reflection, non-conductive areas are almost transparent for
investigation is an integral view of the entire part. Therefore, microwaves. Similar to the Ultrasonic Detection described
defects with a low density could be hidden behind material in section II-A the microwaves are reflected at interfaces
with a high density and thus not be found. Furthermore, the between areas with different permittivity. This method can also
thicker the part is, the more the radiance is absorbed. The be performed in transmission mode what requires a second
resolution of this method can therefore not be defined in antenna and access to both sides of the specimen.
general. In fact, for a quantitative analysis it is necessary Microwave Based Inspection is a new technique and it
to calibrate the measuring system with objects of defined is still under development. According to the manufacturer
dimensions in order to determine the detectable size. it is able to detect defects of about 1 mm at a frequency
X-Ray systems are easily accessible (e.g. at airports or in of 10 GHz. The main advantage is that the analysis works
medical offices) and even bigger parts can be investigated. analog to the established Eddy Current system it is based on.
The most important disadvantages are the comparably low It provides the possibility to determine width and depth of the
optical resolution and the bad sharpness of the edges due inhomogeinities. From the response signal the permittivity of
to scattering. Thus, especially small defects are difficult to the base material and the impurity can be figured out.
find. The detection of adhesion failure between GFRP and the E. Active Thermography
sheath material is not possible.
Every item emits infrared radiance depending on its tem-
C. Computed Tomography perature. This radiance can be visualized with infrared (IR)
Simply expressed Computed Tomography (CT) is a high cameras. In a heat image different colors represent different
resolution X-Ray method [2]. The main differences between temperatures [5]. Active Thermography means, that the surface
CT and conventional X-Ray are that the specimen is inves- of the specimen is heated up by a thermal impulse by a few
tigated circumferentially and not only in two planes. Fur- Kelvin. The thermal energy is conducted from the surface
thermore the pictures obtained are not an integral view of into the bulk of the part. Imperfections exhibit generally a
the entire part but rather axial slices of a definable thickness lower heat conductivity compared to homogeneous material.
(typically a few millimeters). From these two features a three- The heating and the cooling are therefore slower in defects
dimensional picture is calculated with very high contrast and like voids, impurities, and delaminations. Warmer spots on the
spatial resolution compared to X-Ray. Computed Tomography heat image after the thermal impulse are hence an indication
enables to distinguish between materials with differences in for a defect inside the specimen. A variation of Active Ther-
density less than 1% [3]. With CT it is possible to determine mography is to induce friction in delaminations and cracks
the absorption coefficient or the density of the material what by coupling an ultrasonic wave into the specimen. Due to the
makes it feasible to identify a certain material if its absorption friction the part locally heats up what can again be detected
coefficient is known. The costs for a standard CT device are with an infrared camera [6], [7].
typically about 800,000 US$, for high-end equipment higher F. Laser Shearography
than 2,500,000 US$. Laser Shearography is mainly used for the quality inspection
D. Microwave Based Inspection in large components made from composite materials like e.g.
airplane wings. A laser emits a reference beam and an object
The principle of Microwave Based Inspection is based on beam. The superimposition of these two beams results in an
the Eddy Current Test which is well established for non- interference pattern which is recorded with a CCD camera.
destructive testing of metals. For the application to plastic At the beginning of each measurement, an interference
materials the working frequency of the Eddy Current device of picture of the unloaded part is recorded. Afterwards a load
typically 5 MHz is transformed to about 10 GHz. The signal is applied by dynamical excitation, heat or pressure and a
is coupled into the specimen by means of an antenna. Figure 2 second picture is taken. Since the deformation on the surface
shows schematically the functional principle. is different in the vicinity of a defect, inhomogeinities beneath
the surface can be detected. Therefore, Laser Shearography is a
method which potentially should to be able to detect adhesion
failure between silicone and the tube material. It has already
been reported that a defect of about 1 cm2 could be found [8].
The expenses for this test method are comparably high and
it is limited to regions near the surface.
III. R ESULTS
Fig. 2: Microwave Based Inspection [4] In the following results and pictures are presented, which
were detected with four of the methods described above.
Depending on the dielectric properties of the contaminations Initially, there was no intention to evaluate these methods
inside the specimen the emitted waves are reflected in a systematically. Therefore, not each method was tested in order
characteristic manner: Conductive materials lead to a high to assess its ability to detect a certain kind of defect.

537
A. Ultrasonic Detection
As already indicated in section II-A the detection with
ultrasonic is difficult, particularly if the defect is smaller than
1 mm. Adhesion failure between GFRP tube and the housing
material is impossible to find. Flat samples with intentionally
missing adhesion were prepared and sent to two different
manufacturers. Both were not able to detect the defects in (a) Air bubble near the core (b) Air bubble in a shed
a reliable and reproducible way. This is consistent with the Fig. 4: Detectable defects with CT
results described in another work [8].
B. Computed Tomography
1) Insulator: Figure 3 shows the results of the examination 2) Scarf Joint: Figure 5 shows the cross section of an
with computed tomography at three different positions along assembled GFRP tube at the position of the scarf joint.
the axis of an insulator. The pictures were taken in 1 mm
slices.

(a) Flange (b) Flange/Tube (c) Sheath/Tube

Fig. 3: Results of Computed Tomography

Fig. 5: CT picture of a scarf joint
In the left picture (3a) a position in the aluminum flange is
displayed. The radiation is strongly absorbed by the metal, thus There is no visible difference in brightness, i.e. in density
the material appears comparably bright while the surrounding between the base material and the glue used to join the two
of the insulator is totally black1 . partial tubes. Inside the tube an irregular comparably dark
The outer ring in Fig. 3b is again aluminum but the inner structure can be seen. This is excess glue which cures under
ring is the GFRP tube. The latter appears darker in comparison ambient conditions and has therefore a lower density than the
to the metal flange due to its lower density2 . glue inside the joint which cures at a defined pressure.
Finally, Fig. 3c shows the GFRP core surrounded
by the shed material (silicone rubber). Again a signifi- C. X-Ray
cantly darker grey shade can be observed for the silicone In Fig. 6 the insulator also examined in the CT was
((Silicone) 1.1 g/cm3 ). Furthermore, the contrast is very investigated with X-Ray. The defects described in Fig. 4a and
high, i.e. the interface between GFRP and silicone appears 4b should thus appear in the X-Ray pictures.
very sharp.
In Fig. 4 defects are shown which have been created
intentionally in order to find out the limits of the method.
Figure 4a shows a void close to the core of the insulator. The
distance between two large ticks on the right hand scale is
10 mm, what means that the diameter of the void is approx.
1 mm. Due to the large difference in density between the air
inside the defect and the silicone, the contrast is very high
and the edge is displayed sharply. Therefore, it is obvious that
even smaller (down to 0.1 mm) voids could be detected easily
even by unexperienced persons. (a) 0 (b) 90
In the same sample further defects were found in the sheds
(Fig. 4b). The dimensions of these voids are approx. 10 mm Fig. 6: X-Ray of an insulator in two planes
in length and 2 mm in width.
However, even the comparably large defects in the sheds
1 Inmedical diagnostics, the degree of absorption is expressed in terms of can obviously not be identified by X-Ray. One possible reason
Hounsfield Units (HU). Air has -1000 HU, water 0 HU, and metals a couple
of 1000 HU. for this effect is that the largest dimension of the void
2 (Aluminum) 2.6 g/cm3 , (GF RP ) 2.0 g/cm3 points perpendicular to the image plane. That means that the

538
probability to find a defect depends not only on its size but
also on its shape and its orientation.

D. Microwave Based Inspection

1) Scarf Joint: For the investigation of the scarf joint sam-
ples were prepared using less glue than necessary. Figure 7a
shows a simple backlight picture which was realized placing a
lamp inside the tube. The dimensions of the picture are approx.
60 60 mm. A large bright area can be identified with no glue
or significantly less glue than in its surrounding. Although this
is a very simple test it provides results quickly and no special
experience or training is needed for interpretation. However,
the material needs to be at least partially transparent for visible
light and the optical resolution is not evaluated yet.
Fig. 8: Adhesion failure detected by the Microwave method

hollow-core composite insulators reliably. Common techniques

like Ultrasonic Detection often fail due to the non-flat geom-
etry and the complex composition of the materials.
The most suitable method to detect defects in the bulk mate-
rial, i.e. in the GFRP or in the sheath seems to be Computed
Tomography. It provides a high optical resolution (down to
0.1 mm), a sharp contrast, and principally the possibility to
(a) Back light (b) Microwave Inspection
identify extraneous material. The interpretation of the results
Fig. 7: Results of the Microwave based inspection is easy and does not need special training. However, larger
parts (>2.5 m in length and >500 mm in diameter) cannot
The same sample was investigated with Microwave Inspec- not be examined with commercially available equipment. Fur-
tion (cf. Fig. 7b). The detection frequency was 24 GHz in this thermore, the detection of bad adhesion between tube material
case and the measurement was performed in reflection mode. and housing material is not possible, since there is normally
The figure does not show the raw data but a two-dimensional not a real gap with a measurable depth.
false color plot. The dashed rectangle marks the position of In order to detect bad adhesion Microwave Based Inspection
the defect. Although it is not as easy to see as in the back gives first promising results. However, as it is a fairly new
light picture, the shape and the dimensions of the imperfection method, there is not much experience with this technique. The
can also be found with the Microwave method. However, the spatial resolution and the reproducibility have to be evaluated
contrast is not as good as with the back light and the edges in further investigations. Furthermore, the application to an ac-
are not very sharp to identify. tual insulator with a wavy shed profile has not been performed
2) Adhesion Failure: Finally, the Microwave Based In- yet.
spection system was tested with respect to its capability to
R EFERENCES
detect adhesion failure. The specimen already described in
section III-A was used. In accordance to the regular production [1] P. Walter. Detektion und Evaluation von Inhomogenitaten im Volumen und
an inneren Grenzflachen in polymeren Isoliersystemen mittels Ultraschall.
process the GFRP-part was coated with an adhesion agent PhD thesis, RWTH Aachen, 2006.
(primer) before applying the silicone rubber. In the center [2] A. Mentgen. Computertomographie auf dem Vormarsch. Problem und
of the sample a circular spot with diameter approx. 3 cm Losung, 13, 2008.
[3] http://en.wikipedia.org/wiki/X-ray computed tomography.
was covered during the application of the primer in order to [4] J. Hinken and D. Beilken. Microwave Defectoscopy with Extended Eddy
suppress a chemical bonding between the silicone and the Current System. http://www.ndt.net/article/v10n10/hinken16e/hinken16e.
GFRP at this position. However, there was no actual gap htm.
[5] http://en.wikipedia.org/wiki/Thermography.
between the two materials. [6] T. Zweschper, A. Dillenz, G. Riegert, D. Scherling, and G. Busse. Lockin
Figure 8 shows the result as a gray scaled picture. The lattice thermography methods for the NDT of CFRP aircraft components. e-
is an interference pattern and not due to defects. However, in Journal of Non-Destructive Testing, 8(2), 2003. http://www.ndt.net/article/
ecndt02/249/249.htm.
the center of the sample a darker shaded circle with diameter [7] R.B. Mignogna, R.E. Green, E.G. Hennecke, and K.L.Reifsnider. Ther-
3 cm can clearly be identified. mographic investigations of high-power ultrasonic heating in materials.
Ultrasonic, 7:159163, 1981.
IV. C ONCLUSION [8] H. Hillborg. Non-destructive testing of silicone rubber outdoor insulation:
Surge arresters and hollow core insulators. In 2009 World Congress on
Obviously there is no universal non-destructive test method Insulators, Arresters, and Bushings. Insulator News and Market Report,
which is able to detect all considerable imperfections in 2009.

539