During the 1970s, Hydro-Québec (Montréal, Canada) installed nine 250 MVAR (16-kV) synchronous compensators to better control network voltages through dynamic injection of reactive power. These machines were designed with hydrogen (H₂) cooling under a pressure of 30 psig (2 bars relative). Commissioning tests indicated that under design conditions, these machines could be continuously overloaded up to 325 MVAR without excessive overheating.

Over the years, the utility observed several excessive H₂ leakages. For safety reasons, the H₂ pressure and the maximum MVAR loading were reduced on the compensator with the most leakage. Hydro-Québec made temperature rise calculations of critical parts under different loadings and H₂ pressure conditions, and concluded that the 325 MVAR loading should still be possible with the H₂ pressure reduced to 15 psig (1 bar relative). However, field tests revealed that, while the machines withstood the overloading, the overheating at 325 MVAR damaged the bushings (phases and neutrals). While the base machine was essentially Class F insulation, the bushings were Class B.

Designed more than 30 years ago, each of these machines has six identical hydrogen-to-air bushings. Modern bushing designs use polymer insulation materials to increase thermal capabilities, so Hydro-Québec decided to recondition the damaged bushings and increase their current capacity from Class B to Class F. The utility would completely replace the original insulation with polymer insulation.

The following information details the design and the test process to install composite insulated bushings on one of Hydro-Québec's synchronous compensators.

Service Contract Conditions

Hydro-Québec's objective was to avoid a completely new bushing design and focus on replacing the insulation to meet the following service conditions:

• The bushing must carry 12,000 A (325 MVAR) continuously without exceeding the maximum temperature rise allowed for Class F insulation.

• The bushing must withstand a cantilever load as specified in IEC 60137.

• The bushing should withstand a minimum temperature -50°C (-58°F).

• Verify pressure tests (H₂) over the temperature range (-40°C, 60°C [-40°F, 140°F]).

Contractually, the following special requirements were imposed:

• The copper conductor and the flange had to be reused.

• The creepage distance had to remain the same.

• The maximum diameter of the sheds had to be 280 mm (11 inches) to allow use of the same current transformers.

• The connections at both ends should remain the same.

• The original sealing concept had to be the same.

Test Program

Considering the use of H₂ under pressure, security was an important issue and the tightness of the modified bushing was one of the main concerns.

A test program based on IEC 60137 was defined to verify performance, including the following specific service conditions:

• A lightning impulse test for the new polymer insulating material.

• A cantilever test to verify the mechanical stresses on the new material.

• A temperature rise test at rated current (12,000 A) to verify the maximum temperature of the central conductor and the behavior of the new material.

• Tightness tests to verify the sealing at high temperature for normal operation at full summer load and at low temperature during prolonged stop conditions in winter with ambient temperature down to -50°C.

• Thermal cycling tests to verify the design of the interface between the central conductor and the polymer insulation.

For the first three tests, acceptance criteria were based on IEC 60137, but the permissible leakage rate was tight enough for security purposes. The accumulation method was used with a maximum H₂ leakage rate of 5 cc/hour.

The thermal cycling test is not a standardized test, and a special test procedure had to be defined and agreed to by both parties. This included 10 cycles with temperature variations from -40°C ±5°C to 85°C ±5°C (-40°F ±41°F to 185°F ±41°F).

Under the acceptance criteria, the bushing was not to show any deformation or cracking, or any residual stress; the leakage rate had to remain less than 5 cc/hour of H₂; and the partial-discharge measurement after cycling could not exceed 10pC.

Most of the type tests were done at Hydro-Québec Laboratories. Before sending the prototype to IREQ, the manufacturer, …lectro Composite (ECI), checked the tightness with a helium detector (sniffer). ECI used a SF₆ detector for the measurement at IREQ because it could measure the leakage rate at both low and high temperatures. Thus, the acceptance criterion was modified to 1.5cc/hour of SF₆ to take into account the characteristics of the gas.

Test Results, Design Improvements

The test program lasted about 10 months, providing a good insight into the weak points of the new design and led to the following improvements:

Lightning impulse tests performed in air (original BIL 125 kV) at ECI's facility, resulted in external flashover at 120 kV (positive) in the H₂ (pressured part of the bushing); no flashover occurred in negative polarity at 125-kV peak voltage. An additional series of 15 positive impulses at 115-kV peak was then applied without any flashover. ECI decided to accept 115-kV BIL level in air since the maximum operating voltage is only 20 kV (nominal 16 kV). A partial discharge measurement taken after the impulse tests showed no degradation of the insulation (below 10 pC). Nonetheless, ECI decided to modify the position of the ground screen to decrease the dielectric stress in the H₂ (pressured part). This modification was implemented in the second prototype.

Cantilever tests were conducted at IREQ. The cantilever load specified by IEC 60137 was increased from 3.17 kN to 3.93 kN to account for the connector weight (79.4 kg). The load was applied in all four directions successively.

After the test, a visual inspection of the bushing revealed no visible mechanical failure and no permanent deformation. A tightness test performed at 20°C (68°F) revealed an acceptable leakage rate (equivalent to 1.5 cc/hour of H₂) near the bushing's flange.

A temperature rise test was performed at IREQ at a nominal current of 12000 A. It was calculated that, in service, the lower part of the bushing (H₂-pressured) could be subjected to a surrounding H₂ temperature of up to 60°C while the outside ambient could reach 40°C (104°F). It was not possible to reproduce these operating conditions for the temperature rise test. To avoid overstressing the bushing, the conductor temperature could not exceed 150°C (320°F) and the ambient air temperature had to be around 60°C. Thermocouples were installed on the bushing. To reach an ambient air temperature close to 60°C, an enclosure had to be installed around the bushing. The bushing was supplied with current for about 9.5 hours. The ambient temperature was stable at 59°C (138.2°F). The maximum temperatures measured were 138.1°C (280.58°F) inside the central copper conductor and 104.8°C (220.64°F) on the external composite insulation material.

Even if the test conditions were different from service conditions, the results established that the insulating material could support the actual temperature rise conditions since no mechanical problems were discovered.

For the tightness test, the second prototype was used with added seals at both ends of the insulating core. The bushing was installed on a small reservoir and set under SF₆ pressure (45 psig or 3 bars relative). The temperature of the conditioning chamber was lowered to -50°C and maintained for 10 hours.

In this first experiment, the leakage rate at -50°C was greater than anticipated, and major leaks from the tank and bolts did not allow consistent measurements on the bushing. The temperature was increased to -20°C (-4°F) and, using the old soap method, a major leak was detected near the flange. No apparent leak occurred above or below the bushing where seals were added. A second verification at 20°C confirmed a permanent loss of tightness near the flange, which substantiated the weak point at the flange-polymer interface. Since the problem was located, the testing continued. Modifications to the prototype and repetition of the leakage tests were planned after the thermal cycling tests.

The thermal cycling tests were performed at IREQ. Extensometer gauges to measure the displacement and deformation of materials (conductor, flange and polymer) and thermocouples were installed on the bushing. The test program included 10 thermal cycles with temperatures ranging from -40°C ±5°C to +85°C ± 5°C with durations of 6 to 8 hours at these temperatures.

No deformation or cracking was observed on the bushing at the end of these thermal cycles. Measurements by extensometer gauges showed no residual stress. A tightness test at 20°C indicated no tightness degradation. Moreover, the partial-discharge measurements taken after the thermal cycling test showed no degradation of the insulation (<10 pC).

The prototype was modified again to cope with the flange interface tightness problem. The main problem was due to the reusing of an existing flange, which wasn't adequate to fit the polymeric insulation. To solve this problem, a seal was added in the lower part of the flange. New temperature tightness tests were scheduled. The initial leakage rate, as measured at 20°C, showed the modified bushing was tight. The leakage measurements were performed at -35°C (-31°F) and 80°C (176°F) with the accumulation method. Special precautions were taken to remedy the bolt leakage on the tank.

The low-temperature measurement was lowered to -35°C at the rate of 10°C/hour. After 12 hours at -35°C, the H₂ leakage rate was measured at 7.2 × 10^-3 cc/hour, which was well below the limit of 5 cc/hour. The SF₆ pressure measured at -35°C was 33.4 psig (2.3 bars relative). The temperature of the chamber was then increased to 80°C (+10°C/hour). The pressure was limited to 45 psig (3 bars relative).

The H₂ measurements taken after more than 12 hours at +80°C indicated an increased rate (corresponding to 1.55cc/hour) but still lower than the acceptable limit. A new H₂ measurement was taken once the bushing had stabilized at the ambient temperature (20°C) that yielded results of 42 × 10^-3 cc/hour. This last test confirmed that the tightness problems observed previously had been solved.

Manufacturing and Routine Testing

Following the type tests, a retrofit was performed on five additional bushing units. The following routine tests were agreed to with ECI:

• Dry power-frequency voltage withstand test (38 kV, 1 min) followed by measurement of partial-discharge quantity at 17 kV, with an acceptance criteria of 10 pC.

• Measurements of the dielectric dissipation factor (tan d) and capacitance.

• Tightness tests at +20°C.

These tests were performed at ECI's plant. No flashover occurred during the withstand test at 60 Hz, and the partial discharges on the bushings remained below 1 pC.

The test setup used for the routine tightness tests was different from the one used for the type tests. The bushing was installed on a tank filled with helium gas at a pressure of 45 psig (3 bars relative). A vacuum chamber was installed on the tank and the vacuum was created with a pump (0,5 Torr). When vacuum was reached, the leakage rate was measured with a mass spectrometer. Given the test was performed with helium instead of H₂, the acceptable rate was adjusted according to the respective densities of each gas to 4.25 cc/ hour of helium. The helium leakage rate measured for the bushings under routine tests at +20°C ranged from 0.054 cc/hour to 1.8 cc/hour. The test method and acceptance criteria used represented a conservative approach.

Future Development

The reconditioned bushings were subjected to a test program to validate the modified design. The behavior of the conductor interfaces, polymer insulation and flange was verified under different operating conditions. Special attention was paid to the tightness.

Following the initial test results, modifications were made to ensure tightness and a very low partial-discharge intensity level. Six bushings were reconditioned, and the routine tests confirmed they were properly manufactured. These bushings have been reinstalled on one of the synchronous compensators with no problems to date.

Following this experience, it was clear that the existing conductors should not be reused for an upgrading because no assurance could be given as to their condition (smoothness, straightness) before manufacturing. Any delay during manufacture could have a major impact on the availability of the machine. The same holds true for the flanges.

Through this project, Hydro-Québec demonstrated a practical solution to upgrade the bushings on its synchronous compensators. In the meantime, the planning department has reevaluated the requirement for a continuous loading (325 MVAR) for the other machines and decided that a temporary overloading (30 min) is sufficient to meet the actual operating requirements. The inherent thermal inertia of the existing bushings is long enough to withstand this short overload without jeopardizing its integrity. However, if the continuous loading requirement returns, a tested and proven method has been established to upgrade the equipment capacity.

Francine Rochon received the bachelor's of science degree in electrical engineering from Université de Sherbrooke in 1982 and the master's of Science degree in electrical engineering from the …cole Poly-technique de Montréal in 1985. From 1988 to 1991, she worked at IREQ in the high-voltage laboratory as a test engineer for the power transformer department. She joined the Hydro-Québec quality-control department as a test specialist in 1991.

Gilles L. Desilets received the bachelor's and master's of science degrees in electrical engineering from the Université de Sherbrooke in 1966 and 1967, respectively. He has been working with different departments at Hydro-Québec, including IREQ. As a member of the planning department, he was involved with the initial design of the synchronous compensators at Hydro-Québec in the 1970s. He has actively involved with IEC and CIGR… and presently, is responsible for technical support of synchronous and static compensators, and HVDC converter equipment.