1. Introduction
Gas-insulated metal-enclosed switchgear (GIS) has advantages that include high reliability, small ground space requirements, a long maintenance cycle, low environmental impact, flexible construction, good expansibility, etc. GIS is widely used in large-scale hydropower projects, urban high-voltage power grids installed on complex terrains and in narrow areas. As weakness components of GIS, basin insulators are internal insulators in GIS, as shown in
Figure 1, and are used for electrical insulation, isolation of the gas chamber and supporting conductor, and mainly determine the safe and reliable operation of a GIS [
1,
2].
Basin insulators are thoroughly mixed by epoxy resin, alumina filler, and a curing agent in different contents, and they are then manufactured through the casting and curing processes [
3]. Internal stress can be generated in basin insulators, during their manufacture, transportation, installation, and operation. In the manufacturing process, residual stress is caused by chemical shrinkage due to cross-link reactions and thermal shrinkage while cooling down from curing to room temperature [
4]. In addition, internal stress results from the vibration and friction in the transportation; the mechanical force due to unbalanced installation and switching operations; the vibration induced by short-circuit electrodynamic force; the thermal gradient stress due to central conductor operating heating and the different thermal expansion coefficients between the epoxy composite and insert metal in the operation. The internal stress can cause microcracks, which gradually develop to larger sizes, and can lead to gas leakages, partial discharges, insulation flashover, ablation, and the cracking of the whole insulator.
For instance, in 2012, a gas leakage accident happened in a breaker functional unit of a porcelain bushing chamber at the high-voltage side of a transformer in a 220 kV GIS substation in Guangdong Province, China, which was induced by cracking in a basin insulator. This resulted from tangential compressive internal stress, due to an artificial uneven bolt fastening force [
5]. An isolated static side basin insulator, in an isolating switch docked with a bushing, cracked in 2017. The conductor in the switch was inclined to the insulator, resulting in bending tangential compressive stress. The measurement and adjustment of the verticality of the bushing were absent during the bushing installation [
6].
It has been reported that a surface stress distribution on a 252 kV GIS basin insulator was measured during a pressure test [
1] by Shi et al. [
7] using strain gages pasted on an insulator. They measured the surface stress distributions to investigate the developments of the cracking, and to find the vulnerable point on the surface of the insulators, which is useful for improving insulator design and increasing mechanical strength. In this method, it was only the surface stress, but not the internal stress, which was measured. However, the internal stress finally determines the vulnerable points and the mechanical strength of the insulators during the pressure tests [
8]. Nevertheless, there is currently no proposition for measuring internal stress in basin insulators inside a GIS.
To measure internal stress, two kinds of methods could be used: destructive and non-destructive methods. The destructive methods are based on the relaxation of the stress in a material by cutting, breaking, machining, or similar methods, and thus lead to damage to the material. These include the hole-drilling method, ring core method, slot-cutting method, indentation method, layer removal method, etc. The non-destructive methods are based on the relationship between the internal stress and physical parameters, and include five main methods: X-ray diffraction method and neutron diffraction method, based on the evaluation of interplanar spacing in crystal materials [
9]; the magnetic method, based on the magnetostriction effect in ferromagnetic materials [
10]; the photoelasticity method, based on the birefringence exhibited by certain transparent materials [
10]; the ultrasonic method, based on the relationship between stress and ultrasonic wave velocity [
9].
Previous studies have shown that the ultrasonic method could non-destructively measure the internal stress in many composites [
9]. The ultrasonic method for measuring internal stress is based on the acoustoelastic effect, and it is classified according to wave mode into longitudinal waves, shear waves, combinations of shear and longitudinal waves, critically refracted longitudinal waves, Rayleigh waves, etc. [
11].
The ultrasonic stress measurement technique has been mainly developed for the case of metallic materials. Based on ultrasonic longitudinal waves parallel to stress, Jhang et al. [
12] used the phase detection technique to precisely measure the propagation time in high-tension bolts and estimated the clamping forces. However, the effects of ultrasonic system delay were not eliminated in measuring the propagation time. Fukuoka et al. [
13] applied the sing-around method to measure the wave velocity differences through ultrasonic shear waves and determined the residual stress distribution in a patch-welded circular plate made of mild steel. The method of combining shear and longitudinal waves was proposed by Pan et al. [
14] to measure the tensile internal stress along the length of the bolts, made of austenitic stainless steel and low-carbon steel. This method could improve the measurement accuracy, as the effects of temperature and elastic deformation were eliminated. The residual stress in the near surface of patch-welded steel plates was measured by Bray et al. [
15] using critically refracted longitudinal waves, which were excited by a longitudinal transmitter with the first critical incident angle. Duquennoy et al. [
16] developed propagation equations of Rayleigh waves for measuring the residual stress in the lateral surface of aluminum alloy sheets after the quenching process, and the results were consistent with the destructive layer removal method.
At present, there are few investigations regarding the application of the acoustoelastic effect on the internal stress measurements in non-metal materials. Jia et al. [
17] found that the velocity of the longitudinal waves changed linearly with the tensile internal stress in certain polymer materials, such as polycarbonate, polystyrene, polyamide, and polybutylene terephthalate. They used the longitudinal pulse-echo method to measure the ultrasonic velocity based on the time interval between first and second reflected waves at the back surface. Based on the longitudinal pulse-echo method, Xu et al. [
18] observed a linear relationship between the longitudinal wave velocity and the compressive internal stress in polymer-bonded explosives, calculated the acoustoelastic coefficient, and measured the internal stress. In this method, the ultrasonic wave velocity was measured based on the propagation time obtained through a phase correlation algorithm, and the strain was obtained using a scanning acoustic microscope. Wang et al. [
19] proposed an equation that determined the relationship between the stress, in the area beneath the surface, and propagation time of critically refracted longitudinal waves in a carbon-fiber-reinforced plastic composite, and concluded that the propagation time was linearly dependent on the stress. They calculated that the acoustoelastic coefficient was consistent with the experimental results.
There is only a report investigating the acoustoelastic effect in the fiber-reinforced epoxy composite with epoxy matrix, by Santos et al. [
10], who employed the critically refracted longitudinal waves. They showed that the propagation time varied linearly with the tensile internal stress in the fiber direction of 0°, 45°, and 90° and the propagation time was the longest in the 0° direction. However, by using the Lcr waves technique, the stress measurements were only limited to the area beneath the surface of the epoxy composite, and the internal stress inside could not be measured. Additionally, the applications to calculating the stress, using the relationship between the stress and the propagation time, were not demonstrated.
In this study, the tangential compressive internal stress, resulting from mechanical compression in the epoxy composite specimens of GIS basin insulators, was measured based on the acoustoelastic effect. The method based on the longitudinal through-transmission technique, instead of the conventional pulse-echo method, is proposed for the GIS epoxy composite, because there is no second reflected wave at the back surface, due to the high attenuation and long propagation distance of the epoxy composite. The uniformity of the internal stress in cuboid epoxy specimens was verified through finite element simulation, which is important information that previous researchers were unable to make verifications. An internal stress measurement system is developed to investigate the relationship between the internal stress and the velocity of ultrasonic waves vertical to the stress in the epoxy composite, and to calculate the acoustoelastic coefficient. The system delay was measured using two epoxy specimens with two different thicknesses. Additionally, the internal stress measurement errors were calculated and analyzed.
2. Experiment
2.1. Epoxy Composite Specimen
In this study, epoxy composite cuboid specimens were manufactured by a company, with similar materials and manufacturing process to those of 252 kV GIS single-phase basin insulators, and the process is explained as follows. To begin, bisphenol A epoxy resin and alumina filler were mixed with a ratio of 1:4, and then the vacuum membrane degasification was conducted for the first time. Next, the anhydride-type curing agent was added and the mixture was thoroughly blended for 70–80 min; the time duration is long enough to ensure the materials are evenly distributed. The vacuum membrane degasification was conducted for the second time. The liquid mixture was then poured into a preheated casting mold with the speed in the range of 1.5–2 kg/min. Furthermore, two-stage curing was conducted. In the first curing cycle, the temperature was increased to 105 °C in two hours and then maintained for 15 h. In the second curing cycle, the temperature was increased to 155 °C and maintained for 15 h [
20,
21,
22].
There were six cuboid specimens prepared, labeled as #1–6 specimens, with a length (
a) of 45 mm, a width (
b) of 30 mm, a height (
h) of 35 mm, a density of 2.23 g/cm
3, and an elastic modulus of 12.25 GPa. The length (
a) was the same as the thickness of the 126 kV three-phase-in-one-tank type GIS basin insulators. The cuboid specimen has been used for ultrasonic compressive stress measurements in published reports [
18,
23,
24,
25,
26]. If the material of the cuboid specimen is uniform, the internal stress in the specimen will be homogeneous [
23]. The uniformity of the epoxy composite was regulated according to the standard: the difference in filler content should be less than 2% and the difference of density should be less than 0.2 g/cm
3 [
27]. Therefore, the internal stress in the cuboid epoxy specimen is homogeneous. The diameters of alumina filler in the epoxy specimens follow a normal distribution, with 20 μm medium diameter (D50), which is defined as the corresponding particle diameter when the cumulative particle diameter distribution percentage of a sample reaches 50% [
21].
The #1, #2, and #3 specimens were used for investigating the relationship between internal stress and ultrasonic longitudinal wave velocity, as well as for calculating the acoustoelastic coefficient. Based on the obtained acoustoelastic coefficient, the internal stress of the #4, #5, and #6 specimens could be measured.
2.2. Testing System
In this study, an internal stress measurement system using an ultrasonic longitudinal through-transmission method for epoxy composite specimens similar to GIS basin insulators was set up. This system consists of an ultrasonic pulser/receiver, two ultrasonic transducers, an oscilloscope, a compression test machine, a strain test instrument and its strain foils, a computer, epoxy composite specimens, and connecting wires, as shown in
Figure 2.
In the experiment, the ultrasonic signals emitted by a transducer were originally generated by an ultrasonic pulser/receiver (CTS-23, Shantou Institute of Ultrasonic Instruments Co., Ltd., Shantou, China), with a frequency bandwidth of 0.5–3 MHz, an emitting pulse rise time of around 30 ns, and an A-scan type. The working mode of pitch and catch was selected to connect two transducers. The rectification signals, including positive, negative, and bidirectional signals, could be directly taken from the panel of CTS-23. In order to obtain original ultrasonic waveforms without rectification, CTS-23 was improved by connecting the oscilloscope directly with the point before the rectification mode in the electrical circuit inside the CTS-23. Thus, the waveforms that contained the most information could be obtained [
28,
29].
The transmitting and receiving transducers were general longitudinal straight beam transducers (2.5P20, Shantou Institute of Ultrasonic Instruments Co., Ltd., Shantou, China), with a central frequency of 2.5 MHz and a diameter of 20 mm. A QQ9 wire was used to connect the transducers and ultrasonic pulser/receiver. Water was used as the couplant between the transducers and the specimen. Rubber bands were employed to press the transducers against the specimen.
The electrical waveforms were observed with a digital oscilloscope (DPO4104), which was manufactured by Tektronix Co., Ltd. in Beaverton, OR, USA and purchased from Mei Da Ke Digital Technology Co., Ltd. in Guangzhou, China. The sampling frequency was 5 GS/s and the sampling period was 0.2 ns, which allowed for precise measurements of the ultrasonic propagation time in the specimen, as the smallest variation of the propagation time was more than 1 ns. The bandwidth of the oscilloscope was 20 MHz, which was sufficient for measuring the ultrasonic signals emitted at 2.5 MHz and could eliminate the noise with a frequency greater than 20 MHz.
The uniaxial compression tests performed on the specimen were operated by a microcomputer-controlled compression test machine (CMT5105, Shenzhen Xin San Si Material Detecting Co., Ltd., Shenzhen, China), with a capacity of 100 kN and an accuracy level of 1. The accuracy is equal to the ratio of the maximum absolute error and capacity; therefore, in the range of 100 kN, the accuracy class of 1 indicates that the maximum absolute error is 1 kN [
30]. The uniaxial compressive loading, gradually increased by an equal-stress gradient with a speed of 0.5 mm/min, was applied on the specimen (stress was the ratio of force
F to cross-sectional area and the cross-section was 45 mm (a) × 30 mm (b)). The propagation time was measured at each 5 MPa interval from 0 to 70 MPa, and thus, 15 times in total. As the cracks occurred in the specimen under approximately 130 MPa stress, the maximum allowable stress reached 70 MPa, corresponding to 94.5 kN, which was within the range of the compression test machine.
A high-performance static tester (JMTS-116, Jing Ming Technology Co., Ltd., Yangzhou, China) with strain foils was used to measure the strain of the specimens under stress along the
z-direction. The strain foils were used to collect the strain data of
εy in the
y-direction and
εz in the
z-direction, shown in
Figure 3. The strain accuracy of ±10
−6 was sufficient in this study, as the strain variation was around 10
−4. The dimensions of specimens were measured using a Mitutoyo micrometer with an error of 0.001 mm.
2.3. Finite Element Simulation
A finite element simulation was employed to investigate the uniformity of internal stress in specimens during the compression experiments in the range of 0–70 MPa, using COMSOL software (COMSOL 5.4). This software was manufactured by COMSOL Co., Ltd. in Stockholm, Switzerland, and purchased from Kang Mo Shu Er Software Technology Co., Ltd. in Shanghai, China. The simulation model includes a three-dimensional cuboid epoxy specimen, which has the same dimensions and material properties, including the density, Young’s modulus, and Poisson’s ratio, as those of the specimens used in the experiments.
5. Conclusions
In this study, an internal stress measurement system using the proposed ultrasonic longitudinal through-transmission method was developed to measure the compressive internal stress in the epoxy composite specimens of GIS based on the acoustoelastic effect. The following conclusions can be drawn.
(1) The uniformity of the internal stress in cuboid epoxy composites can be verified through finite element simulation, in the case that the material is uniform.
(2) The ultrasonic longitudinal through-transmission method can be applied for the epoxy composite specimens of GIS with the property of the high attenuation and the long propagation distance. There is a good linear relation between the stress and ultrasonic longitudinal wave velocity, with a high linear correlation coefficient, in GIS epoxy composites. The average acoustoelastic coefficient of the GIS epoxy composites using the longitudinal waves vertical to the stress was found to be 4.556 × 10−5/MPa.
(3) The ultrasonic system delay should be considered in measuring the ultrasonic wave propagation time if the through-transmission method is used, and the delay can be measured using two specimens with different thicknesses.
(4) The internal stress has been calculated using the proposed method. The absolute errors of the ultrasonic internal stress measurements are less than 12.397 MPa, and the relative errors are large in the stress state of 10–15 MPa, with the largest value at 85.788% and less than 19.876% in the stress state of 40–70 MPa.
(5) The reason for the measurement errors include the small ratio of the ultrasonic wave velocity variation to the stress variation, random errors, instrument measurement errors, and dimension deviations during the manufacturing process of epoxy specimens.
This research shows that the ultrasonic internal stress measurement method for GIS epoxy composites is feasible, providing a basis for the research to measure the internal stress in basin insulators. The technique can be used as quality control for new insulators in factories and to analyze insulators after they go into disuse.