Abstract: This paper briefly describes the fault situation of surge arresters used for CVT compensation reactor protection, analyzes the cause of the damage, and concludes that it was caused by excessive voltage drop of the surge arrester when measuring the dielectric loss of the CVT. Based on this, a reasonable selection of CVT test methods is proposed. Keywords: CVT; surge arrester; fault; compensation reactor; test During the pre-testing work, many surge arresters used for CVT protection of compensation reactors were found to be severely damaged. The causes and hazards of surge arrester damage are analyzed, and rectification measures are formulated to ensure the safe operation of CVT. 1 Fault Situation During the pre-testing of the 220kV VII busbar CVT at a 220kV substation, the testers (using a 2500V megohmmeter) found that the insulation resistance at the X terminal of the A-phase intermediate transformer was zero. The next day, it was found that during the pre-testing of the 220kV busbar CVT at another substation, the insulation resistance at the X terminal of the three-phase intermediate transformer was zero for phases A and B, and only 1MΩ for phase C. The internal wiring of the CVT is shown in Figure 1, and the equivalent circuit is shown in Figure 2. [align=center]Figure 1 Internal Wiring Diagram of CVT 2 Equivalent Circuit of CVT L—Compensating Reactor MOA—Surge Arrester J—Carrier Device T—Intermediate Transformer Z—Damper[/align] From the CVT wiring diagram, it can be seen that the surge arrester is directly grounded at its end. When measuring the insulation resistance at terminal X, the grounding at terminal X is disconnected. If the surge arrester is damaged, current can flow through the surge arrester to ground. Therefore, it was initially determined that the surge arrester breakdown was the cause. Subsequently, the manufacturer's technicians confirmed this judgment and replaced the surge arrester. Simultaneously, the surge arrester's tail end was changed from being directly grounded internally to being connected in parallel to terminal X, grounded externally. 2. Cause Analysis It was learned that the surge arrester is a protective measure taken to suppress the ferroresonance inside the CVT. The compensating reactor connected in parallel with the surge arrester has linear impedance. When the CVT is operating normally, the voltage drop across this reactor will vary depending on the size of the secondary load, but generally does not exceed 700V. If ferroresonance occurs, the intermediate transformer core saturates, and the current flowing through the primary winding and compensating reactor of the intermediate transformer will increase rapidly. This causes the voltage drop across the reactor to exceed the operating voltage of the surge arrester, causing the surge arrester to conduct and disconnecting the reactor. By changing the circuit parameters, the ferroresonance can be eliminated. There are three possible causes for surge arrester damage. First, repeated ferroresonance during CVT operation causes a voltage rise across the compensating reactor, resulting in the surge arrester being subjected to a higher voltage and thus damaging it. However, if ferroresonance is the cause, the probability of simultaneous resonance in a set of three phases is very small. When measuring the insulation resistance of the primary winding of the intermediate transformer, the secondary winding is open-circuited, and the X terminal is ungrounded, while the tail end of the surge arrester is effectively grounded inside the tank. Its equivalent schematic diagram is shown in Figure 3. Figure 3: CVT equivalent schematic diagram during the test. When measuring with a 2500V megohmmeter at the X terminal, a DC voltage of 2500V is applied to the X terminal. Since the reactor is inductive, the voltage U at the start of the surge arrester is approximately equal to 2500V. The rated voltage of this metal oxide surge arrester is only 800V, which may damage the arrester during testing. Therefore, a known good surge arrester of the same model was tested, and the specific test data is shown in Table 1. The test results show that using a megohmmeter will not damage the surge arrester. This is because, although the voltage of a 2500V megohmmeter is sufficient to keep the surge arrester in a conducting state, its output current is very limited, only 2mA, which is insufficient to accumulate enough energy to break down the valve plate. When performing tests to measure the dielectric loss and capacitance of the voltage divider capacitor, conventional and self-excited methods are often used for pre-testing. The following analyzes the impact of different wiring test methods on the surge arrester. The conventional method measures the capacitance and dielectric loss of C1 and C2 connected in series. A positive connection is used, with 10kV applied to the U terminal and a signal taken from the δ terminal (approximately grounded). The secondary winding is open-circuited, the X terminal is ungrounded, and the tail end of the surge arrester is grounded, as shown in Figure 3. When 10kV is applied to terminal U, C1 and C2 are connected in series, and the voltage at terminal Ua ≈ C1U/(C1+C2) ≈ 3kV. The secondary winding of the intermediate transformer is open-circuited, and the tail end X of the reactor is floating and ungrounded. Given that the surge arrester has a relatively large impedance, the primary winding of the intermediate transformer can be approximated as short-circuited, and the voltage is almost entirely applied to the surge arrester. Its equivalent circuit diagram is shown in Figure 4. Figure 4: Equivalent circuit diagram of the surge arrester during the test. In the figure: voltage U' is the voltage at terminal A, approximately 3 kV; ZX is the equivalent impedance of the primary winding when the secondary winding of the intermediate transformer is open-circuited, with a small resistance; Z is the equivalent impedance of the surge arrester, with a relatively large resistance. Because the no-load impedance ZX and U of the primary winding are constant, the voltage at the start-up terminal L of the surge arrester during the dielectric loss test is determined by the impedance Z of the surge arrester. When Z→∞, UL=U≈3kV. This subjects the surge arrester to a power frequency voltage close to 3kV, which has a significant destructive effect on the protection of the surge arrester. The self-excited method is used for measurement. When measuring the dielectric loss and capacitance of voltage divider capacitors C1 and C2 using the self-excited method, a voltage is applied to the auxiliary winding dadn. Utilizing the electromagnetic induction of the intermediate transformer, a 3kV high voltage is induced on the primary side of the intermediate transformer. This voltage is the test voltage applied to the voltage divider capacitors C1 and C2. At this time, terminal X is grounded, consistent with the wiring method during CVT operation. There are two reasons for the damage to the surge arrester used for compensating reactor protection: first, due to the CVT structure, the surge arrester is always in the circuit during the test, thus repeatedly bearing the test high voltage, leading to breakdown; second, the conventional wiring method used in the test causes the surge arrester to bear high voltage. It is recommended to strengthen communication between the manufacturer and testing personnel, ensuring that the manufactured equipment meets the needs of on-site testing. Furthermore, before conducting CVT tests, testing personnel should carefully analyze the equipment structure and verify different test wiring methods to ensure safe and efficient testing.