1

ANALYSES OF COMPOSITE INSULATORS WITH CRIMPED END-FITTINGS: PART II - SUITABLE CRIMPING CONDITIONS

M. Kumosa, D Armentrout, L. Kumosa, Y. Han and S. H. Carpenter

Center for Advanced Materials and Structures

Department of Engineering University of Denver

2450 S. Gaylord St., Denver, Colorado 80208 Abstract In this research the mechanical behavior of composite suspension insulators with crimped end-fittings manufactured by NGK has been evaluated. Two issues have been addressed in this study. Insulators crimped with the standard, high, reversed and constant high stress conditions have been analyzed for their crimping deformations, and the two and three dimensional compression profiles for the four different crimping conditions have been determined. Second, a series of mechanical tests have been performed by subjecting the insulators to excessive tensile axial loads until failure. Acoustic emission (AE) was also monitored during the tests. It has been found that the shape of the compression profiles is dependent on the manufacturing crimping process. The magnitudes of compression for the standard, high, reversed and constant high stress conditions are almost the same, with the distribution of compression along the fitting being noticeably different for each stress condition. The results from the pull-out tests clearly demonstrate that the crimping process can significantly effect the mechanical strength of the insulators. The insulators with the standard controlled stress conditions failed by rod pull-out whereas the other crimping conditions resulted in rod fracture. It has been shown in this study that the mechanical strength of crimped insulators can be optimized to avoid both crushing of the glass reinforced polymer (GRP) rods due to crimping and rod fracture under excessive tensile loads. I. Introduction A description of the design of composite insulators with crimped end-fitting has been presented in Ref. 1. In these insulators, the metal end-fittings are radially compressed onto the ends of the rod and the bonding between the GRP rod and the fittings is purely mechanical in nature. It was also shown in Ref. 1 that the mechanical behavior of crimped insulators is very strongly dependent on the magnitude of applied compression and compression profiles. The previous (Ref. 1) was predominantly numerical. Two and three-dimensional fully non-linear finite element models of crimped insulator ends were constructed and the mechanical behavior of the insulators was numerically simulated. It was shown that non-uniform crimping conditions generated significantly higher tangential and radial stresses in the rod in comparison with uniform crimping. Most importantly it was presented in Ref. 1 that under axial tension crimped composite insulators developed large tensile stresses in the rod at the top of the fitting. The magnitude of the tensile stresses in this location depended very strongly on the magnitude

Published version can be found in Composites Science & Technology, Vol. 62, No. 9 (2002) pp. 1209-1221.

2

of applied compression and compression profiles. To verify the numerical data presented in Ref. 1, a series of mechanical tensile tests on crimped composite insulators with various crimping profiles was performed in this study. II. Experimental Procedures and Results Several crimped composite insulators were prepared for the crimping analysis by NGK Insulators, Ltd. The insulators were crimped with the following four crimping conditions: standard stress condition, high stress condition, reversed stress condition and constant high stress condition. The crimping conditions are schematically illustrated in Figure 1. Points A and B in Figure 1 indicate the locations in the beginning (A, upper surface of the fitting) and end (B, lower surface) of the fitting. The compression profiles on the rod surface for the four different types of insulators were determined. Subsequently, a series of mechanical tensile tests were performed to establish relationships between the type of compression profile and the type of insulator failure. 2.1 Determination of compression profiles on the rod surface The major difficulty associated with the investigation of different crimping techniques on the long and short-term properties of composite insulators is the proper evaluation of the crimping deformations on the rod surface inside the fitting. Essentially, there are three ways of measuring the deformation on the rod surface, Ur

rod. This can be accomplished by using an ultrasonic technique [2-8]. If the crimping deformations on the rod surface are relatively large (larger than 0.1 mm) the application of the ultrasonic technique can be very successful. However, if the compression on the rod surface is small (below 0.1 mm) the error associated with the ultrasonic measurements can be significant. This is especially true in the case of composite suspension insulators with small rod diameters (d =16 mm). The ultrasonic measurements can also be affected by the geometrical properties of the GRP rods. If the rod is either bent or not circular in nature the error can be almost as large as the value of compression. The rod surface compression can also be estimated by sectioning an insulator along the fitting and measuring the compression either by optical or scanning electron microscopy (SEM) [6,8]. Using this technique, a compression profile can be determined by measuring the diameters of the GRP rod inside the fitting (under compression) and comparing them with the diameter of the rod before compression was applied. However, this method cannot be used with a large degree of accuracy if the internal surface of the fitting is threaded [8]. The third method that can be used to determine the compression applied during manufacturing is based on the measurements of the external diameter of the fitting [1,8]. If the external diameters of the fitting, along its length and circumference, are measured in the uncompressed and compressed areas the actual crimping deformations applied during the manufacturing process to the fitting can be estimated to a large degree of accuracy using equation (1) presented below Ur

rod = Ure − G (1)

3

where Ur

e is the applied crimping on the end-fitting, Urrod is the effective crimping on the

rod surface, and G is the gap between the rod surface and the internal surface of the fitting. To determine the actual compression on the rod surface, the gap between the internal surface of the fitting and the rod surface before crimping must be known. 2.1.1 Analysis of crimping deformations using an ultrasonic technique At first, the manual contact pulse-echo ultrasonic technique [2-8] was used to determine the extent of radial compression exerted on the glass reinforced polymer (GRP) rods of four randomly selected insulators from NGK with the standard, high, reversed and constant high stress conditions. Before measuring the internal compression, the external surfaces of the fittings were machined off in order to remove the surface plastic indentations generated during the crimping process. The external surfaces of the end-fittings were discretized into 16 lines in the circumferential direction (φ) and 13 lines in the axial direction (z). This created a grid with 221 data points. The distance between the points was 0.2 inch (approximately 5 mm). Then, an ultrasonic transducer (Parametric 26DL plus, Dual Element Transducer, 10MHz) with a diameter of 5 mm was acoustically coupled to each of the 221 grid points, and the interface depth x, was measured. The internal radius of the compressed rods, rin, was measured as rin = ref - x (2) where ref is the measured external radius of the end fitting. ref was measured mechanically. From the outer radius of the uncompressed GRP rod, r, the extent of radial compression, Ur, of the rods was determined from Ur = r -rin. (3) The compression profiles u(φ,z) of the insulators were obtained with an accuracy of 0.025 mm. The radius of the rods (before crimping) was provided by NGK and was approximately 8 mm. The first set of the compression profiles were obtained on the fittings which were not accurately machined. Significant misalignment of the surface of the fittings with respect to the rod axis was found. Therefore, the shape of the profiles was significantly affected by the machining error. Therefore, another set of insulators with four different crimping deformations were machined and analyzed for their crimping deformations using the ultrasonic technique. Major efforts were made to minimize the machining error. The misalignment of the fittings with respect to the rod surface outside the fitting was found to be smaller than 0.025 mm. However, the quality of the profiles was still highly questionable. Additional problems with the alignment of the rods were discovered. It was

4

found that a slight distortion of the rods, during the pultrusion process, significantly effected the measurements of the compression profiles. The profiles could be corrected to account for the machining error. However, the error caused by the distortion of the rod cross section cannot be corrected since the rod misalignment along an insulator is not constant, and can be different for each insulator. As an example, two three-dimensional compression profiles from two insulators with the standard and high stress conditions are shown in Figures 2a and 2b, respectively. 2.1.2 Analyses of rod surface compression using equation (1) Since the ultrasonic measurements of the crimping profiles showed not to be accurate for the insulators considered in this study, additional experiments were performed to determine the actual compression profiles in the insulators investigated. The compression of the rod surface inside the fitting after crimping was determined by measuring the diameter of the crimped fitting in several locations along the fitting and using equation (1). Seven randomly selected insulators for each crimping condition were evaluated. The two-dimensional axial (along the fitting) profiles for the insulators with the standard stress, high stress, reversed stress and constant high stress conditions are shown in Figures 3 (a-d). In addition, the average crimping deformations (average out of seven) along the fitting for the four crimping conditions are shown in Figure 4. It can be seen from the diagrams presented in Figures 3 and 4 that the profiles are different for each crimping condition. However, for each crimping condition small variations in the compression profiles can be observed. For the standard stress condition (see Figures 3a and 4), the amount of compression of the rod surface in the areas close to the top surface of the fitting (point A) is lower than the compression deep inside the fitting. The radial compression in the beginning of the fitting is approximately 0.08 mm whereas the compression inside the fitting is about 0.11 mm. The profiles deep inside the fitting are essentially uniform. However, in the initial part of the profile the compression increases almost linearly along the fitting. For the high stress condition (see Figures 3b and 4), the profiles are similar to the profiles for the standard stress conditions. However, the magnitude of the compression close to the rod (in the beginning of the fitting) is slightly higher (0.10 mm) for the high stress conditions than for the standard stress conditions. It can be notice in Figures 3c and 4 that the profiles for the reversed stress conditions are different. The maximum radial compression near the top end of the fitting is approximately 0.097 mm whereas the compression near its lower end is less (0.08 mm). For the constant high stress condition, the profiles are almost uniform with the compression slightly increasing along the fitting. Similar to the case of the reverse stress condition, the radial compression on the rod surface in the beginning of the fitting is higher than for the standard and high stress conditions and is about 0.10 mm. The magnitudes of compression (the average compression on the rod surface inside the fittings) were also determined and found to be almost the same for the four types of crimping conditions considered. In particular, the average magnitudes of compression for

5

the standard, high, reversed and constant high stress conditions were 0.0998 mm, 0.103 mm, 0.0988 mm and 0.109 mm, respectively. The values of the magnitudes of compression determined using equation (1) are entirely different than the magnitudes of compression from the ultrasonically determined three dimensional compression profiles. This clearly indicated that the compression profiles obtained using the ultrasonic technique were not correct. It can be seen from the above diagrams that the average compression of the rod surface is almost the same for the four different crimping conditions. However, the compression profiles are slightly different. The lowest compression near the top end of the fitting was found for the standard stress condition and was lower than for the other four crimping conditions. It will be shown in the sections below that these very small variations in the compression profiles can have tremendous effects on the mechanical performance of the insulators when subjected to large tensile mechanical loads. 2.2 Mechanical Tensile Tests The pull-out strength of NGK insulators with the four different crimping conditions was investigated when deformed in a tensile mode. Three insulators, from each group, were tested with very reproducible results. The insulators were tested in a standard MTS deformation machine. The samples were deformed in stroke control at a rate of 0.254 mm/min. The acoustic emission generated during deformation was also monitored using a resonant (150 kHz) transducer placed near the end where pull-out was expected. A 40 dB preamplifier was used in the AE experiments. 2.2.1 Mechanical and acoustic emission test results The load versus displacement and the sums of acoustic emission events versus displacement curves for one set of insulators with different crimping profiles are shown in Figures 5(a-d). In addition, the four load-displacement curves and the maximum loads at failure are presented in Figures 6(a) and 6(b), respectively. Results from the other two sets of insulators are essentially equivalent except for one insulator with the reversed stress conditions which failed in mixed mode (pull-out and rod fracture almost simultaneously). An examination of Figures 5(a) and 6(a) shows that the load curve for the standard stress condition is the most unique. It starts out with a linear region, bends slightly, comes to a maximum, has a small drop and continues to maintain a significant load for a long period of time. The other three crimping conditions give somewhat similar load/displacement curves (see Figures 5 (b-c) and Figure 6(a)) showing a linear region, a slight bend over to a maximum and then a series of rapid load drops with a loss of load carrying capability. It is important to notice that the maximum loads were not strongly dependent on the crimping conditions. Figure 6(b) shows the scatter in the maximum load for the three test runs for each crimping condition. Notice in both cases that the scatter is very small.

6

The loads and displacements at failure for the insulators subjected to the pull-out tests are presented in Table 1. In addition, the loads and displacements at the bend of the load/displacement diagrams are also listed in this table. Acoustic emission generated during the deformation of the isolators was measured and found to be unique for each method of crimping (see Figures 5 (a-c) and Figure 7). The different crimping conditions produced significantly different acoustic emission. The insulators with the standard stress condition produced the lowest amount of acoustic emission (see Figure 7). The emission showed a steady increase as the rod was pulled from the fitting. Undoubtedly, the acoustic emission was produced from the friction between the composite rod and the fitting. For the insulator with the high stress condition, the acoustic emission showed a rapid increase at a first load drop and as the load continued to fall. Unfortunately, the acoustic emission measurements are not complete. At the load drop, the sudden shock was sufficient to dislodge the transducer from the sample and further acoustic emission was not measured. The acoustic emission generated for the reversed stress condition is approximately five times greater than that generated by the constant high stress condition. However, in both cases acoustic emission shows a rapid increase as the load begins to drop after reaching a maximum. The loss of load was very sharp and rapid for the reversed stress condition with a corresponding rapid increase in the acoustic emission. The insulator with the constant high stress condition had a slower and less rapid loss of load, and hence a less rapid increase in the acoustic emission. Clearly, in all but the standard stress condition the acoustic emission was primarily associated with the rod fracture and was composed of events from fiber fractures and matrix splitting. 2.2.2 Failure morphology The most defining features observed for the four different types of crimping were the fracture surfaces of the insulators subjected to the pull-out tests. Figure 8 shows a set of samples after being pulled to failure. The four sets of insulators failed in the following manner: Set 1. Standard stress condition: three insulators failed by pull-out. Set 2. High stress condition: three insulators failure by rod fracture Set 3. Reversed stress condition: two insulators failed by rod fracture and one by rod fracture and pull-out (mixed mode) Set 4. Constant high stress condition: three insulators failed by rod fracture The insulators with the standard stress conditions failed by the insulator rod simply being pulling out of the end-fitting. No visible damage to the rod was observed. However, traces of white powder inside the fitting and on the rod surface inside the fittings were detected. The failure of the insulators crimped with the high stress conditions was characterized by a large crack in the rod inside the fittings, with considerable splitting in

7

the rod. The reversed stress crimping was also characterized by cracks in the rod well down in the end grip. However, the cracks were smaller and much more splitting and pull-out was observed. The constant high stress conditions gave by far the most splitting and pullout of the four samples. In Figures 9 (a-c) the morphologies of the fracture surfaces for the insulators with the high, reversed and constant high stress conditions are shown. 2.2.3 Pull-out tests up to the maximum loads To examine the differences in the observed failure modes, four insulators with the standard stress, high stress, reversed stress and constant high stress conditions were tested up to the first significant drop on their load/displacement curves. After the load drops, the specimens were immediately unloaded. Subsequently, the fittings were removed and the surface of the rods inside the fittings was carefully examined. The load reduction for the insulator with the standard stress condition was associated with the onset of macroscopic sliding of the rod out of the fitting. Since the internal surface of the fittings was slightly threaded, the load reduction was caused by a sudden movement of the rod inside the fitting under tension. No damage to the rod was found, except small traces of white powder indicating that the fine thread on the surface of the fitting did not create any significant damage to the rod during crimping and pull-out. In the remaining three insulators, the sudden load drops coincided with the formation of very short transverse surface cracks in the GRP rods inside the fitting in the regions in the beginning of the fittings immediately followed by axial debondings along the rod. The damage detected in the four insulators is shown in Figures 10(a-d) with and without the fittings. In Figures 10(b-d) the arrows indicate the locations of the surface transverse cracks in the rods responsible for the drops in load. III. Finite Element Modeling In Ref. 1 a comprehensive fully non-linear finite element analysis of crimped composite insulators has been presented. The insulators were modeled as two and three-dimensional structures (see Figure 11) assuming either uniformly or linearly distributed crimping profiles. Two types of linear profiles were considered. In type 1, the maximum compression was applied in the beginning of the fitting (point A) whereas in type 2, the maximum of the linearly distributed compression was applied at the bottom of the end-fitting (point B). Major differences in the numerical load/displacement curves from the numerical pull-out and the internal stress distributions were found depending on the compression profile assumed. It was found that large tensile stresses develop in the rod along the fibers when the rod is pulled out of the fitting for the uniform and linear type 1 crimping profiles. The location of the tensile stress concentration in the rod was near the upper surface of the fitting (point A). To evaluate the effect of axial tensile loads on the stress concentrations in the GRP rod inside the fitting, additional finite element computations were performed for the three different crimping profiles (uniform, linear type 1 and linear type 2). The internal stresses in the fittings were determined for an axial load of 150kN which represents the average

8

load at failure for the insulators tested in this study (see Figure 6(b)). The stress distributions in the three fittings are essentially the same as in Ref.1 (Figures 17, 20 and 21 in Ref. 1) except for their magnitudes. The maximum normal axial stresses in the rod σz determined inside the fitting for an applied tensile load of 150 kN are: 1. Uniform crimping: 922 MPa 2. Type 1 linear crimping: 1058 MPa 3. Type 2 linear crimping: 846 MPa The locations of the tensile stress concentrations in the GRP rod for the uniform and linear type 1 crimping profiles are almost the same and occur near the top surface of the fitting for the uniform and type 1 linear crimping cases. For type 2 linear crimping, the tensile stresses are concentrated in the rod, deep inside the fitting. The lowest tensile stresses in the rod are for the type 2 crimping profile. It is clear from the results presented above that by reducing the compression on the rod surface in the beginning of the fitting, the lowest tensile stress concentrations in the rod are generated during pull-out. Since the insulators with the standard stress conditions failed by rod pull-out, it was important to establish the internal stress distributions in their end-fittings. By doing this the optimum stress conditions in crimped insulators could be established. Therefore, an additional finite element analysis of the crimped insulator with the standard stress condition was performed. The average crimping deformations, as shown in Figure 4, were prescribed on the rod and the stress distributions inside the end-fitting were determined just after crimping and during pull-out at 150 kN. These two stress distributions are shown in Figures 12a (after crimping) and Figure 12 b (during pull-out). It can be seen in Figure 12a that the reduction in the compression in the beginning of the fitting resulted in a reduction of the radial and tangential stresses in the GRP rod. The other stress components were either zero or very small, as indicated in Figure 12a. When tensile forces were applied, the internal stress distributions changed dramatically. The radial and tangential stresses in the rod were substantially reduced in the areas close to point A, similar to the observations made in Ref. 1. Deep inside the fitting the reduction of these two stress components was almost insignificant. It can also be observed that the normal stress σz in the rod near the upper surface of the fitting exhibited a noticeable stress concentration. The shear stress σrz increased in comparison with the pure crimping case (without applied tension) and was found to be less than 50 MPa deep inside the fitting. However, in the location where the normal stress σz was concentrated the σrz

stress component in the rod was very small. IV. Discussion The biggest problem in the failure analysis of crimped composite insulators is the determination of the compression profiles caused by crimping. Without the proper determination of the actual crimping deformations on the rod surface due to crimping, the mechanical strength properties of crimped insulators cannot be accurately evaluated. It has been shown in this research that the ultrasonic technique cannot be used for this

9

purpose if the crimping deformations are small in crimped insulators with small rods. The accuracy of the compression profile measurements using this technique can be very strongly affected by: A. Misalignment of the surface of the fittings, caused by machining, with respect to the long axis of the rod. This effect seems to be less pronounced in crimped insulators with large rods [2-5]. B. Misalignment of the GRP rod inside and outside the fitting. If the rod is bent and its cross sections are not circular the shape of the compression profiles can be significantly affected especially if the crimping magnitudes are small. C. Accuracy of the ultrasonic measurements. The accuracy of the ultrasonic measurements performed in this study was approximately 0.025 mm which was approximately 25% of the surface deformation of the rods inside the fitting of the insulators considered in this project. This experimental error could be significantly smaller for crimped composite suspension insulators with much larger crimping deformations [7,8] without having a noticeable effect on compression profiles. The results from the pull-out experiments clearly indicate that the modes of failure of the crimped insulators from NGK are dependent on the crimping conditions generated during the manufacturing process. The insulators with the standard stress conditions failed by rod pull-out. The other three crimping conditions, high stress, reversed stress and constant high stress resulted in the fracture of the GRP rod. It was shown in this research that these four different crimping conditions generated slightly different compression profiles. The magnitudes of compression on the rod surfaces were almost the same with only small differences in the distribution of compression inside the fitting. Despite this, the modes of failure were significantly different. It appears that the type of insulator failure, either pull-out or rod fracture, depends on the shape of the compression profiles associated with the four different crimping conditions and the internal stress distributions in the rod caused by the four different crimping conditions. The strength of the NGK insulators, however, seems to be less effected by the crimping conditions. The maximum loads determined from the pull-out experiments for the insulators are essentially the same with only small variations in the loads at failure. Taking into account the finite element results presented in this study and in Ref. 1, two important conclusions can be drawn. First, the uniform crimping conditions create stresses in the rods which are predominantly compressive with uniformly distributed radial and tangential stresses in the rod inside the fitting. These stress distributions change dramatically if axial tensile loads are applied. If a crimped insulator with uniform compression is subjected to tension, significant tensile stress concentrations appear in the rod near the top surface of the fitting. If the stress concentration is higher than the longitudinal tensile strength of the composite rod material, the rod should fracture in this location. Second, it has been shown that the linear crimping conditions create radial and tangential stresses in the rod due to crimping which are not uniformly distributed. For both type 1 and type 2 crimping conditions, the radial and tangential stresses in the rod

10

are much higher than the stresses for the uniform crimping case. Therefore, in order to design an insulator free from any intralaminar damage due to excessive crimping and at the same time avoiding rod fracture during pull-out, a compromise must be reached as far as the type of compression profile is concerned. It appears that the best combination of crimping conditions is the combined case of uniform crimping and type 2 linear crimping. If we look at the data presented in Figures 3a and 4 the compression profiles for the standard stress conditions are essentially a combination of type 2 linear crimping and uniform crimping. Therefore, this type of crimping must generate the stresses in the rod which will avoid rod crushing (the uniform crimping part of the profile) and rod fracture during pull-out (the linear part of the compression profile). For the other crimping conditions, the profiles are similar therefore damage to the rod during crimping is not an issue. The fact that compression for the standard stress condition near the top end of the fitting is slightly lower in comparison with the other three profiles resulted in the failure of the insulator by pull out, instead of rod fracture. For the other three crimping conditions, the values of compression in the beginning the fitting were slightly higher; therefore, the pull-out tests resulted in rod fracture. The longitudinal tensile strength of most unidirectional glass/polymer composites is in the range from 800 to 1000 MPa. If this stress is not exceeded, the rod should pull out without fracture. However, if the tensile stress concentration in the GRP rod near the top surface of the fitting is greater than the longitudinal tensile strength of the GRP rod then the rod will fracture in the direction perpendicular to the fibers. It can be seen in Figure 12b that the insulator with the standard stress conditions developed a noticeable normal stress concentration near the top surface of the fitting when pulled at 150 kN. This stress however was smaller than the longitudinal tensile strength of the GRP rod. Therefore, the insulators with the standard stress conditions failed by rod pull-out. A small increase in the crimping compression near the top surface of the hardware (cases with the high, reversed and constant stress conditions) resulted in the failure of the insulators by rod fracture. The normal stress concentrations in the beginning of the fitting were responsible for the initiation of the short transverse cracks in the rods as shown in Figures 10 (b-d). The initiation of the transverse cracks was immediately followed by the formation of axial debonds (cracks along the rod). Since the initial transverse cracks suddenly reduced the cross sections of the rods, the normal tensile stresses (σz) were rapidly increased, exceeding the longitudinal tensile strength of the composite. This resulted in rod fracture. It has been shown in Ref. 1 that the normal stress concentrations in uniformly crimped insulators increase in proportion to the magnitude of applied compression for a given applied tensile load. Therefore, the initiation of the transverse cracks could have occurred at much lower tensile loads if the values of compression had been significantly higher. Conclusions 1. The mechanical tests results obtained from the pull-out tests clearly show that the failure process of the NGK insulators is strongly dependent on the crimping process. The insulators with the standard stress condition of crimping failed by pull-out. The other

11

crimping conditions resulted in the mechanical failure of the rods. Very little scatter has been observed in the loads at failure for all the insulators investigated in this project. It was also shown that the failures by pull-out or by rod fracture occurred almost at the same load level. This means that by proper manufacturing the strength of the insulators can be maximized with the insulators failing by pull-out for the loads which are almost equal to the loads at rod fracture. 2. When crimped composite insulators are subjected to large tensile axial loads, the internal stress distributions inside the end fittings are significantly altered in comparison with the stresses just after crimping. For the uniform and linear type 1 crimping profiles, significant tensile stress concentrations appear in the GRP rod inside the fitting close to the upper surface of the fitting. These stress concentrations can be responsible for the fracture of the GRP rod inside the fitting during pull-out if the tensile stresses in the rod exceed the longitudinal tensile strength of the composite rod material. However, if the crimping profiles are linear, close to the type 2 crimping profile, the tensile stresses are less concentrated. In this case, the region with the maximum tensile stress in the rod will be located deep inside the fitting, close to the lower end of the fitting. 3. In order to maximized the strength properties of crimped composite insulators, for both crimping and pull-out, a crimping profile should be a combination of the uniform and linear type 2 crimping cases. The compression in the rod in the beginning of the fitting should be linearly distributed with the smallest compression near the top surface of the fitting. The second part of the profile inside the fitting should be uniform. However, the value of the smallest compression on the surface of the rod is also important. If this value is too large, the rod will fracture under axial tension. 4. The composite insulators from NGK investigated in this project exhibit the highest possible strength properties. Their compression profiles have been designed to minimize the possibility of internal cracking in the GRP rod due to crimping. At the same time, by carefully optimizing the profiles for the insulators with the standard stress condition, the failure of the rod during pull-out has been avoided. The standard stress condition generates a compression profile which is a combination of the uniform and liner type 2 crimping profiles. This agrees very well with the conclusions from the numerical part of this study. References 1. M. Kumosa, Y. Han and L. Kumosa, Analyses of Composite Insulators with Crimped End-Fittings: Part I - Non Linear Finite Element Computations, submitted for publication to Composites Science and Technology.

2. M. Kumosa, Analytical and Experimental Studies of Substation NCIs, Final

Report to the Bonneville Power Administration, Oregon Graduate Institute of Science & Technology, Portland, Oregon, 1994.

12

3. A. Bansal, A. Schubert, M. V. Balakrishnan and M. Kumosa, Finite Element Analysis of Substation Composite Insulators, Composites Science and Technology, Vol. 55 (1995) pp. 375-389.

4. A. Bansal, Finite Element Simulation and Mechanical Characterization of

Composite Insulators, Ph.D. Dissertation, Oregon Graduate Institute of Science & Technology, Portland, Oregon, 1996.

5. A. Bansal and M. Kumosa, Mechanical Evaluation of Axially Loaded Composite

Insulators with Crimped End-Fittings, Journal of Composite Materials, Vol. 31, No 20 (1997) pp. 2074-2104 .

6. M. Kumosa, Y. Han, S.H. Carpenter, D. Armentrout and L. Kumosa, Suitable Crimping Conditions in Composite Suspension High Voltage Insulators, Final Report to NGK Insulators, Ltd, 1998.

7. M. Kumosa et al, Fracture Analysis of Composite Insulators, Final Report to the Electric Power Research Institute, Department of Engineering, University of Denver, Denver, 1998. 8. L. Kumosa, An Evaluation of Available Methods for the Determination of Crimping Profiles, Internal Report, University of Denver, 1999. Acknowledgments This research was supported by NGK Insulators, Ltd. Additional funds were also provided by the Electric Power Research Institute under contracts #W08019-21 and EP-P2971/C1399. The authors are grateful to Mr. Ishiwari of NGK and Dr. J. Stringer of EPRI for their support of this study.

13

Sample # Crimping

condition Maximum

load (KN)

Displacement @ Maximum load

(mm)

Load @ bend in load curve (KN)

Displacement @ bend in load curve

(mm) set #1

#6 standard 148 7.54 111 4.57 #13 high stress 142 6.99 114 4.57 #24 reverse stress 144 7.01 110 4.45 #34 constant stress 146 7.82 109 4.45

set #2 #3 standard 147 7.34 110 4.45

#12 high stress 148 6.02 101 3.81 #22 reverse stress 142 6.58 116 4.83 #32 constant stress 151 8.56 112 4.57

set #3 #2 standard 148 7.26 111 4.57

#14 high stress 155 8.03 109 4.57 #23 reverse stress 153 7.72 109 4.45 #33 constant stress 141 6.73 109 4.45

Table 1. Loads and displacements at the bent on the load/displacement curves and at failure for the insulators subjected to pull-out.

14

Standard Stress Condition

Reversed Stress Condition

High Stress Condition

Constant High Stress Condition

GRP Rod GRP Rod

GRP Rod GRP Rod

End Fitting

End Fitting End Fitting

End Fitting

Crimping Stress Crimping Stress

Crimping Stress Crimping Stress

A

A

A

B

B

B

B

A

Figure 1: Four crimping conditions analyzed in this research.

15

A

B

(a)

A

B

(b)

Figure 2: Examples of three dimensional compression profiles from the ultrasonic measurements; standard

stress condition (a), and high stress condition (b).

16

0.06

0.07

0.08

0.09

0.10

0.11

0.12

0.13

0.14

0.15

0 10 20 30 40 50 60

Position along the fitting [mm]

Rad

ial c

ompr

essi

on [m

m]

Test 1Test 2Test 3Test 4Test 5Test 6Test 7

A

B

Figure 3a: Radial compression profiles for seven insulators with the standard stress conditions.

17

0.06

0.07

0.08

0.09

0.10

0.11

0.12

0.13

0.14

0.15

0 10 20 30 40 50 60 70Position along the fitting [mm]

Rad

ial c

ompr

essi

on [m

m]

Test 1Test 2Test 3Test 4Test 5Test 6Test 7

A

B

Figure 3b: Radial compression profiles for seven insulators with the high stress conditions.

18

0.06

0.07

0.08

0.09

0.10

0.11

0.12

0.13

0.14

0.15

0 10 20 30 40 50 60Position along the fitting [mm]

Rad

ial c

ompr

essi

on [m

m]

Test 1Test 2Test 3Test 4Test 5Test 6Test 7

A

B

Figure 3c: Radial compression profiles for seven insulators with the reversed stress conditions.

19

0.06

0.07

0.08

0.09

0.10

0.11

0.12

0.13

0.14

0.15

0 10 20 30 40 50 60Position along the fitting [mm]

Rad

ial c

ompr

essi

on [m

m]

.Test 1Test 2Test 3Test 4Test 5Test 6Test 7

A B

Figure 3d: Radial compression profiles for seven insulators with the constant high stress conditions.

20

0.06

0.07

0.08

0.09

0.10

0.11

0.12

0.13

0.14

0.15

0 10 20 30 40 50 60Position along the fitting [mm]

Ave

rage

radi

al c

ompr

essi

on [m

m]

.

Standard StressHigh StressReversed StressConstant High Stress

A

B

Figure 4: Average compression profiles for insulators with the standard stress, high stress, reversed stress and constant high stress conditions.

21

0

20

40

60

80

100

120

140

160

0 2 4 6 8 10 12Displacement [mm]

App

lied

Load

[kN

]

0

300

600

900

1200

1500

Sum

of A

cous

tic E

mis

sion

Eve

nts

Applied LoadAE

Figure 5a: Load/displacement curve and the acoustic emission data for an insulator with the standard stress condition.

22

0

20

40

60

80

100

120

140

160

0 2 4 6 8 10 12Displacement [mm]

App

lied

Load

[kN

]

0

1000

2000

3000

4000

5000

6000

Sum

of A

cous

tic E

mis

sion

Eve

nts

Applied LoadAE

Figure 5b: Load/displacement curve and the acoustic emission data for an insulator with the high stress condition.

23

0

20

40

60

80

100

120

140

160

0 2 4 6 8 10 12Displacement [mm]

App

lied

Load

[kN

]

0

10000

20000

30000

40000

50000

60000

Sum

of A

cous

tic E

mis

sion

Eve

nts

Applied LoadAE

Figure 5c: Load/displacement curve and the acoustic emission data for an insulator with the reversed stress condition.

24

0

20

40

60

80

100

120

140

160

0 2 4 6 8 10 12Displacement [mm]

App

lied

Load

[kN

]

0

2000

4000

6000

8000

10000

12000

14000

16000

18000

Sum

of A

cous

tic E

mis

sion

Eve

nts

Applied LoadAE

Figure 5d: Load/displacement curve and the acoustic emission data for an insulator with the constant high stress condition.

25

0

20

40

60

80

100

120

140

160

0 2 4 6 8 10 12Displacement [mm]

App

lied

Load

[kN

] StandardHighReversedConstant

Figure 6a: Mechanical test data from the pull-out tests; load/displacement curves for one set of insulators with different crimping stress conditions.

26

0

20

40

60

80

100

120

140

160

180

Standard Constant HighStress

High Stress Reversed

Max

imum

Loa

d [k

N]

Figure 6b: Mechanical test data from the pull-out tests; maximum loads from all pull-out tests.

27

0

10000

20000

30000

40000

50000

60000

0 2 4 6 8 10 12Displacement [mm]

Sum

of A

cous

tic E

mis

sion

Eve

nts

Reversed Stress

Constant High Stress

High Stress

Standard Stress

Figure 7: Acoustic emission results from the pull-out tests for four insulators with different crimping conditions.

28

Figure 8: Insulators with four different crimping conditions after the pull-out tests.

Constant High Stress

Reversed Stress

High Stress

Standard Stress

29

Figu

re 9

a: F

ract

ure

surf

ace

of th

e G

RP

rod

for a

n in

sula

tor w

ith th

e hi

gh s

tres

s co

nditi

ons.

30

Figure 9b: Fracture surface of the GRP rod for an insulator with the reversed stress conditions.

31

Figure 9c: Fracture surface of the GRP rod for an insulator with the constant high stress conditions.

32

Figure 10a: Damage in the insulators after the first load drop shown with and without the end-fittings for the standard stress condition.

33

Figure 10b: Damage in the insulators after the first load drop shown with and without the end-fittings for the high stress condition.

34

Figure 10c: Damage in the insulators after the first load drop shown with and without the end-fittings for the reversed stress condition.

35

Figure 10d: Damage in the insulators after the first load drop shown with and without the end-fittings for the constant high stress condition

36

Figure 11a. Finite element representation of the insulator end; 3-D model.

Figure 11b. Finite element representation of the insulator end; 2-D axisymetric model.

37

-400

-300

-200

-100

0

100

0 20 40 60 80 100

z (mm)

Nor

mal

and

She

ar S

tress

es [M

Pa]

Fitting Rod

Figure 12a: Stresses along the rod surface after crimping for an insulator with the standard stress conditions.

σrr = σθθ

σrz

σz = σθz = σrθ = 0

38

-400

-200

0

200

400

600

800

0 20 40 60 80 100

z (mm)

Nor

mal

and

She

ar S

tress

es [M

Pa]

Fitting Rod

Figure 12b: Stresses along rod surface during pull-out at 150 kN for an insulator with the standard stress conditions.

σrr = σθθ

σrz

σθz = σrθ = 0

σz