DESIGN OF ROTATING WHEEL DIP TEST SYSTEM FOR STANDARD TRACKING AND EROSION TESTING OF POLYMERIC INSULATORS

The 20 th International Symposium on High Voltage Engineering, Buenos Aires, Argentina, August 27 September 01, 2017 DESIGN OF ROTATING WHEEL DIP TEST SYSTEM FOR STANDARD TRACKING AND EROSION TESTING OF POLYMERIC INSULATORS J. V. Klüss 1* and J. Hamilton 1 1 Mississippi State University, 406 Hardy Road, Mississippi State, MS 39762, USA *Email: joni@ece.msstate.edu Abstract: Surface tracking and erosion the irreversible degradation on the surface of an insulator can lead to significant increase in surface conduction ultimately leading to failure of an insulator. Polymeric insulation (non-ceramic insulators, NCI) have numerous advantages over traditional ceramic insulators, including superior hydrophobicity correlating to reduced surface leakage current. However, polymers are prone to degradation and aging resulting from electrical, chemical and environmental stress. Tracking and erosion tests expose test samples to artificial harsh environments to verify the suitability of insulator designs and materials (detect weaknesses) by applying continuous, cyclic or combined stresses. Various forms of tracking and erosion testing exist incline plane (IEC 60587), spray (method 1) or rotating wheel dip tests (RWDT, method 2). Numerous standards such as CSA C310-09, ANSI/NEMA C29.13 and IEEE Std. C37.41, and IEC 62730 define the required conditions and procedures for rotating wheel dip tests saline solution, voltage stress, exposure cycles, recovery periods, etc. However, instructions for how to build such setups are not given. The Mississippi State University High Voltage Laboratory is constructing four independent setups for different insulator types and voltage ratings. Simplicity of design and cost effectiveness are priorities. This paper presents the components, configurations, and alternatives for constructing a cost effective stand-alone RWDT system for tracking and erosion testing of NCIs as well as compares the procedures and acceptance criteria presented in various standards. 1 INTRODUCTION In some parts of the world, contamination is the dominant design criterion, dictating the design of line or substation insulation. Contamination flashover requires sufficient deposit (contaminant) on the insulator surface as well as moisture. The contamination by itself does not create a problem, but the mixture of contamination and moisture produces an electrically conducting layer on the insulator surface. Depending on the insulator design, current may concentrate in localized areas resulting in increased surface heating and the formation of a dry band (localized dry region). These dry areas interrupt the current thus forming a voltage across them. Once voltage increases sufficiently high, flashover occurs across the bands and eventually bridges across the entire insulator. [2]. Contamination flashover can be mitigated by improving insulator designs as well as material properties. Ceramic insulators are made from arc resistance material, with excellent internal strength, and immunity to ultraviolet (UV) radiation. For ceramics, the main parameter of contamination performance is creepage (leakage) distance high leakage distance improves performance. Polymer insulators (composites) are not as arc resistant as ceramics and are affect by UV radiation. However, better contamination performance, lighter weight, high flexibility, less maintenance, resistance to vandalism, and easy manufacturability to long lengths makes them a desirable option when selecting insulators for outdoor applications [2]. Ceramics have a well-known performance history and are considered reliable and long lasting. Failures are typically related to incorrect manufacturing process, poor raw materials, or improper application. Non-ceramic insulators (NCIs) were first introduced in the 1960s and experienced a range of different problems related to inappropriate housing materials and manufacturing resulting in water penetration, hardware separation, chalking, tracking, erosion and UV damage. Technological advances in the 1980s resolved many of these issues and NCIs have seen extensive application within the last decades as replacements for suspension and post type ceramic insulators [1]. The first NCIs were made from epoxy and experienced problems outdoors and in contaminated environments. Now, the inner core is typically made from fiberglass providing mechanical strength and is surrounded by a polymer housing and sheds. One of the key characteristics affecting the contamination performance of NCIs is surface hydrophobicity (ability to repel water). As explained earlier, contamination flashover involves dry band arcing which develops due to heating and evaporation of conducting liquid paths. For surfaces with high

hydrophobicity, water forms individual beads or droplets. This formation of droplets inhibits leakage current (discontinuous current path) and the associated dry band arcing [1]. Exposure to surface discharge, corona, and chemicals reduces the hydrophobicity of polymer surfaces. Surface tracking refers to the process that forms irreversible degradation by formation of conductive paths (tracks) starting and developing on the surface of an insulating material [5]. Erosion refers to irreversible and non-conducting degradation of the surface of the insulator that occurs by loss of material [4]. As the outer layer s resistance to tracking and erosion under contaminated and wet conditions is of particular importance, most industry standards include tracking and erosion tests. 2 TRACKING AND EROSION IEC TR 62730 points out that tracking and erosion tests are not considered ageing tests as the test does not replicate real life degradation conditions nor does it accelerate conditions to provide a lifeequivalent test in a short time [4]. Instead, tracking and erosion tests are better described as screening tests used to detect weaknesses related to inadequate materials and designs. The transition from healthy insulation to failure is unpredictable and occurs rapidly. The time and speed of this transition is dependent on a number of parameters insulator material, design, and operating environment. The silicone surface (housing) on modern composite insulators has the ability to recover its highly hydrophobic surface state after it has been lost as a consequence of discharge. This is due to high and low molecular chains constantly breaking down and recombining. This restoration process is expected to last for the expected life of the insulator [1]. The shed material is significant in defining tracking and erosion resistance, but the shape of the sheds also has an influence in defining the leakage path. Thus, the tracking and erosion test should assess all of the aforementioned contributors. The test provides a general indication of the quality of the design and material with respect to the harsh (but not extreme) environment [4]. A number of different standardized tests for evaluating the erosion resistance of outdoor insulators exist, including e.g., the inclined plane tracking and erosion test (IPT), salt-fog test, and the rotating wheel dip test (RWDT). The rotating wheel test has several alternative configurations spray or dip. This paper provides the details for constructing the setup for dipping the insulator in a saline solution, hereafter referred to as the RWDT setup ( Method 2 in [6] or Tracking Wheel # 2 in [5]). 2.1 Standard Procedures The original tracking wheel test was introduced in the Canadian Electrical Association Light Weight Insulator Working Group CEA LWIWG-01 in 1991 [4]. The procedure called for applied voltage stress equivalent to 35 V/mm of leakage distance. The insulator sample was dipped into a NaCl solution of 1.40 g/l of water and allowed to drip. This process was repeated 30,000 without interruptions. To successfully meet the acceptance criteria, surface tracking, erosion to the core, and puncturing of the shed or housing was not allowed. In addition, every unit needed to pass a steep front impulse test and a power frequency voltage test. In later versions, a 24 hour rest period was introduced where the dip tank is empty (to allow for the hydrophobicity recovery mentioned earlier). Although the descriptions slightly differ between the different IEEE, ANSI/NEMA, IEC, CSA standards [2 5] as well as various national level standards (e.g., PN-EN 62217 [7]), the overall procedure today remains the same. A number of samples are selected for the test (two pairs of identical insulators as described in IEC or three samples as described in CSA, IEEE, and ANSI/NEMA). Each insulator is exposed to 30,000 cycles where one cycle consists of 4 periods (positions) energized period, cooling period, dip period, and drip period (Figure 1) Figure 1: The four positions (periods) within a complete cycle in the tracking and erosion test (rotating wheel dip test, RWDT). During the dip period, the insulators sample is submerged in a saline solution consisting of deionized water with 1.40 g/l (± 0.06) During the energizing period, power frequency voltage across the sample shall be:

35 V per millimeter of leakage distance (according to ANSI/NEMA C29.13-2013 for composite deadends and CSA310-09 for polymeric cutouts) No less than 58% of the device s maximum rated voltage (according to ANSI/NEMA C37.41-2016 for fuses and accessories) Creepage distance (in mm) divided by 28.6 (according to IEC TR 62730 for polymeric insulators). For all of the discussed standards, a maximum voltage drop of 5% is allowed when the test circuit is loaded with a resistive current of 250 ma. The cycle time is approximately 200 seconds. IEC and CSA present fixed 40 second stationary periods in each position with an approximate 8 second transition time between positions. ANSI/NEMA and IEEE defines the stationary periods as no less than 80% of the cycle time which is given as 200 ± 25 seconds. All, except EIC require a 24 hour rest period without the saline solution every 4 days. IEC and CSA (but not IEEE and ANSI/NEMA) requires the solution to be replaced weekly. Allowable maintenance periods vary between standards. IEEE and ANSI/NEMA allows up to 8 hours with the full cycle remaining valid (not included in the cumulative test time). IEC and CSA permit 1 hour interruptions as well as one 60 hour longer interruption. An additional testing time of three times the duration of this longer interruption needs to be added to the total duration. Thus, only considering the required 30,000 cycles with an approximate duration of 200 seconds as well as the mandatory 24 hour rest every 4 days, the approximate duration for the tracking and erosion test is 2083 hours (86.6 days, not including maintenance interruptions). As such, the construction of multiple test setups allowing for concurrent testing of different insulator types and classes (for example, prototype testing of a new product) considerably reduces the cumulative test duration. 2.2 Standard Acceptance Criteria The major differences between various standards is apparent in the acceptance criteria. IEC simply requires a resistance measurement after completing the full 30,000 cycles 1. Resistance is measured with a minimum of 1 kv DC with the 1 In fact, IEC has selected the 1000 hour salt fog test as the only standardized test in IEC 62217 applicable to all insulator types in place of the tracking wheel and 5000h (multiple stress) test. probes 5-10 mm apart. Measured resistance has to be greater than 2 MΩ. For composite insulators, erosion is not allowed to reach the core, nor is any puncturing of the shed or housing permitted. IEEE C37.41 requires measured resistance to be greater than 3 MΩ and CSA 310.9 requires R > 1 MΩ. In addition to resistance measurements, both CSA and IEEE require a steep front impulse (1000 kv/µs) test as well as a power frequency test. The power frequency test compares the average flashover voltage with that obtained from an unconditioned reference sample, as well as subjects the conditioned samples to 80% of the reference flashover voltage for 30 minutes. Both standards also record the temperature immediately after removal of test voltage (10 C maximum allowable temperature rise with respect to reference samples). In addition to the aforementioned tests, IEEE C37.41 requires a lightning impulse withstand test at 80% of the sample s rating. IEEE C37.41 specifically states that pre-boiling the tested samples is not required. Similar to the other standards, the post-cycling evaluation in ANSI/NEMA C29.13-2012 calls for a steep-front impulse voltage test and power frequency test as detailed in Clause 7.1 of that standard. However, Clause 7.1 refers to the Water Penetration Test which includes preconditioning of the samples prior to performing the steep front impulse and power frequency test. The preconditioning includes, measuring the hardness of the sample sheds before and after a 100 hour boil in 0.1% NaCl water mixture. 3 DEVELOPED RWDT SETUP As was mentioned earlier, multiple testing setups allowing for concurrent testing of different insulator types and classes significantly reduces total test duration when assessing a large batch of insulators. Although the procedure is the same for the various test samples, the design must allow for customization catering to varying test voltages, different insulator dimensions and shapes, as well as alternative cycling periods and transition durations. The discussed standards provide the testing procedures and acceptance criteria, but they do not define the test setup needed to stress the samples. A number of authors [8, 9] have presented experimental data obtained using RWTD setups but do not go into detail concerning their construction. Some construction details are available in [10-12] but involve programming of the rotating wheel (software). [7] presents an impressive assembly, but may not be the most economical construction. Mississippi State University High Voltage Laboratory (hereafter referred to as MSU) is constructing four RWDT setups. As such, the main criteria are focused on cost-effectiveness,

simplicity, and assembly from readily available resources. The goal is to allow for simultaneous testing of four difference classes or types of insulators. The different insulator types include, suspension type insulators (deadends), support (post) insulators, pin insulators, and fuse cutouts. The different types, as well as different classes within a particular type of insulator, all have different leakage distances. IEC TR 62730 for polymeric insulators restricts the creepage distance to 500 800 mm for the samples to be tested. IEEE, CSA, ANSI/NEMA for cutouts and deadends do not include such a creepage range. The minimum leakage distances for distribution deadend insulators range from 355mm (class DS- 15) to 1190 mm (DS-69 kv) [6]. For 1190 mm leakage distance, the IEC would define the required voltage stress to be 1190/28.6 = 41.61 kv. ANSI/NEMA calculates the corresponding value in a slightly different manner, but produces a similar voltage equal to 41.65 kv. The typical expected cutouts to be tested at MSU are within 38 kv ratings. ANSI/NEMA C37.41-2016 requires no less than 58% of the device s maximum rated voltage, which corresponds to 22 kv. Although ANSI/NEMA C29.17-2013 for composite line post insulators requires the samples to be tested according ANSI C29.11 Clause 7.3, which is a spray method (not RDWT) [14], if one were to perform the RDWT on post insulators, class 57-36 samples with creepage distance equal to 1346 mm would require a voltage stress of approximately 47 kv. Overall, the typical samples to be tested would require a maximum AC voltage stress of 50 kv. In general, approximate clearances for AC-voltage (50/60 Hz) range 100 300 kvrms/m where 150 kvrms/m could be used as an average approximation [13]. Thus, for the expected maximum stress of 50 kv, minimum clearance between grounded and energized components in the constructed test setup must be approximately 30 cm. As the construction is highly inhomogeneous, this minimum clearance value should be increased further to avoid flashover. Simulations could be used to assess the detailed geometry or alternatively a sufficiently large margin added to the theoretical clearances. Field grading can also be used as needed. Along with creepage distance and the corresponding test voltage level, the height of the samples also dictates minimum dimensions for the test assembly. The tested samples must be completely submerged in the saline solution during the dip stage. For example, class DS-69 composite deadend insulators have a section length of 750 ± 75 mm. This defines the required depth of the saline solution tank. If regular off-the shelf containers are used (typically bathtub shaped), products with depths of 750 mm also have considerable width, thus increasing the total volume of the tank. As such, to minimize the volume of needed saline solution, a narrow rectangular tank should be used. One should nevertheless be careful not to make the tank too narrow to limit the available space for e.g., cutouts with the fuses and disconnector assemblies. As evident, the test sample characteristics (shape and size) are dominating variables in defining the dimension of the tracking and erosion test setup. 3.1 Components The design can be optimized using customized solutions but this option typically increases costs. As such, one priority in the MSU design was to implements as many standard products and commercially available materials as possible along with resources already available at the MSU High Voltage Laboratory. The constructed setup is shown in Figure 2. Figure 2: MSU design. A.) controller, B.) voltage and current display, C.) HV source and voltage divider, D.) HV electrode, E.) voltage control (variac), F.) cooling fan, G.) motor. 3.1.1 Frame Since different samples require different clearances, the frame is constructed from 1.5 x 1.5 T-slot aluminium extrusions allowing for easy adjustability while also providing the necessary mechanical support. Furthermore, the construct can be easily disassembled for storage when not in use. The T-slots also allows for convenient routing of wires.

3.1.2 Rotating Wheel Like the frame, the mounting wheel was designed to allow for adjustability. The custom designed wheel and mounts are milled from aluminium and allow for the use of different test specimens. The wheel is grounded and separated from the motor using a fiberglass (insulated) shaft to avoid any current flowing into the motor during the energization stage. 3.1.3 Contact Electrode Different samples (cutouts, deadends, etc.) require different contacts. The spherical electrode shown in Figure 3 is suspended using fiberglass guy strain insulators and has several degrees of freedom and convenient height adjustment allowing for a smooth but solid contact when the sample transitions into position. 3.1.5 Control Board A custom microcontroller board was initially designed for controlling the rotation of the wheel [15]. Upon further consideration, the control board was replaced by an off-the-shelf pulse controller in efforts to simplify the build. The pulse controller is bundled with a control software that allows the user to setup simple control algorithms, making the required cycling protocol easier to replicate. The protocol needs only to be initiated by the software e.g., using a laptop, after which the computer system can be disconnected and used to program the cycles for the other three RWDT systems. Thus, upon initiation of the cycling process, the system is genuinely a stand-alone test assembly. Each control cycle is determined by taking the steps per revolution, which is set on the stepper motor driver, and dividing by the number of test specimens attached to the wheel. The start and stop functions are also handled by the pulse controller. The pulse controller will hold the ON/OFF pin high, 5VDC, when the cycle is intended to progress and will pull the ON/OFF pin low, 0VDC, when the cycle ends. The original microcontroller board will be further developed and used for handling the safety interlock system, safety lighting, and other I/O needs. 3.1.6 Saline Solution Tank As mentioned earlier, a custom fit tank can be manufactured for the submersion of the samples to minimize the required volume of saline content. To reduce expenses, an off-the-shelf high-density resin water tank was selected for the MSU design. The composite material resists corrosion from the saline solution used in testing. The tanks of all four tracking and erosion setups are connected in series with a pump to drain the solution when needed. Figure 3: Flexible contact between sample and HV electrode. 3.1.4 Motor and Driver A commercially available high torque stepper motor with sufficient torque to rotate the wheel with the multitude of different test specimens was selected. A stepper motor allows for precise control of the rotations abiding to the varying standard transition time and stationary periods within a cycle. The MSU design utilizes a NEMA 34, 1700 oz-in bipolar torque, 2.5 A stepper motor selected based on its high torque rating enabling the testing of larger test specimens. To ensure compatibility, the stepper motor driver was selected based on the motor manufacturer suggestions. 3.1.7 Voltage Supply Expenses related to the procurement of four individual voltage supplies were averted by utilizing existing test transformers available at the MSU laboratory. Variable autotransformers (also commercial products) are used to regulate the voltage supply to the test samples. Capacitive voltage dividers are used to measure the applied voltage (suitable high voltage probes can also be utilized). Pearson coils monitor current. Digital displays present current and voltage values. Hand held multi-meters or other data acquisition units displaying values can also be used to observe the test voltage or alternatively digitizers (more costly) can be used to record waveforms. The HV components are all situated on top of the frame

behind Plexiglas for additional safety as well as minimising the dimensions of the construction (floor space). All of the transformers selected to supply adequate voltage stress within the varying sample classes are single phase dry type voltage transformers and have fairly similar dimensions and weight thus allowing all to be installed in a similar manner at the top of the construction behind the touch barrier. 3.2 Budget The total cost of the presented RWDT test system is approximately $2500 USD per assembly. This includes the frame, stepper motor, stepper motor driver, controller, control system power supplies, and machining of the mounting wheel. The cost of the entire assembly (without prior available resources) is approximately $4000 USD. The voltage supply and measurement system form the major cost associated with the construction of a tracking and erosion test assembly. These costs were minimized by using devices already available at the MSU laboratory. When designing the tracking and erosion test fixture, the need for making the design as economical as possible was a large factor as the need for testing multiple samples in parallel is required to keep the test program on schedule. 4 CONCLUSIONS The requirements and acceptance criteria may vary within different standard but the procedure for tracking and erosion is rather consistent. In order to find defective specimens, samples are exposed to four stages energization, cooling, dipping, and dripping. This cycle is repeated 30,000 times corresponding to a test duration of approximately 3 months. Multiple simultaneous test assemblies would considerably reduce the total testing time for a large batch of different samples. To minimize construction costs, this paper presents a practical self-contained design assembled from common easily available components. The design, including the controller circuit is freely available for anybody to use for the construction of a tracking and erosion test setup suitable for their own needs. The first test trials with the presented design are pending upon arrival of test samples. REFERENCES [1] E. Kuffel, W.S. Zaengl, and J. Kuffel, High Voltage Engineering: Fundamentals, 2 nd Ed., Newnes, 2000, pp. 522-528. [2] A. R. Hileman, Insulation Coordination for Power Systems Marcel Dekker, 1999, pp. 701-728. [3] IEEE Std. C37.41, Design Tests for High- Voltage (>1000 V) Fuses and Accessories, 2016. [4] IEC/TR 62730, Technical Report HV polymeric insulators for indoor and outdoor use tracking and erosion testing by wheel test and 5000h test, 2012. [5] C310-09, Canadian Standards Association Distribution class polymeric cutouts, 2010. [6] ANSI/NEMA C29.13, American National Standard for Insulators Composite- Distribution Deadend Type, 2012. [7] T. Samborski, A. Zbrowksi, and S. Koziol, The system for testing the resistance to the surface discharge and erosion of the polymeric insulators, Prezeglad Elektrotechniczny, Vol. 88, No. 3a/2012, pp. 187-190, 2012. [8] H. Ullrich, and S. Gubanski, Accelerated Laboratory Ageing of Model Insulator Samples with Semiconducting Glazes, International Symposium on Electrical Insulating Materials, Himeji, Japan, November 2001, pp. 274-277. [9] R. Tripathi, S. Grzybowksi, and R. Ward, Electrical Degradation of 15 kv Composite Insulator under Accelerated Aging Conditions, Electrical Insulation Conference, Ottawa, Ontario, Canada, June, 2013. [10] A. Muncivi, P. Sarkar, and A. Haddad, Tracking Wheel Test Facilities, 44th International Universities Power Engineering Conference (UPEC), Glasgow, United Kingdom, September, 2009. [11] A.S. Krzma, M. Albano, A. Haddad, Comparative Performance of 11 kv Silicone Rubber Polymeric Insulators under Rotating Wheel Dip Test, 49th International Universities Power Engineering Conference (UPEC), Cluj- Napoca, Romania, September, 2014. [12] A. Gazzola, Tracking Wheel Test for DC Polymeric Insulators, Thesis, Universita Degli Studi di Padova, 2011. [13] J. Kluss, and A. Dickerson, The Effect of Inhomogeneous Ambient Conditions on Air Clearances in High Voltage Laboratories, 35 th Electrical Insulation Conference (EIC), Baltimore, USA, 2017. [14] ANSI/NEMA C29.11, Composite Insulators Test Methods, 2012. [15] J. Hamilton, Tracking and Erosion Build, URL: http://www.ece.msstate.edu/high-voltagelab/tracking-and-erosion/, last accessed: June 14, 2017.