Cable Insulation Assessment Using Partial Discharge Techniques

Generator health is one of the single most critical elements of a power-producing operation's ability to maintain a reliable revenue stream. These generators typically operate at medium voltage levels; thus, a high percentage of overall failures are attributed to insulation defects, as is often the case in any medium voltage equipment.

Statistics from IEEE and EPRI studies indicate that approximately 37% of generator failures can be attributed to stator insulation failure. Stator insulation deterioration can be tracked by regular partial discharge (PD) testing or continuous monitoring methods. The importance of monitoring the generator's insulation for PD activity — along with monitoring other common mode failure components such as bearing condition — is a widely accepted practice in North America. This article will briefly address the general application of multiple generator partial discharge sensors and then discuss a recent event where the sensors predicted an impending failure.

This article describes the real-life, step-by-step approach of an online insulation condition assessment survey at a facility with a small distribution system.

A couple of weeks later, the customer called and explained that the general contractor on site for an unrelated chiller upgrade project stated that the fault would have certainly damaged the customer-owned line-side cables connecting to the utility and recommended replacing the entire MV cable system. It was agreed that a better plan was to first evaluate the cable condition before spending $500,000 on a system replacement. Recommendations included online and offline cable assessment tests to best evaluate the cable system (see sidebar, “Primary Advantages of On- Line Cable Partial Discharge Testing”). These methods identify different cable system failure mechanisms; however, conducting both types of tests would yield the best data for a proper evaluation. The customer agreed to proceed with online PD testing and would follow up with off-line testing at a later date if no PD was detected.

We were all holding back our grins when we merged onto Pacific Coast Highway. PCH runs tight along the California coast for over 600 miles from the Oregon border to Mexico and is perhaps the most beautiful coastal route in the U.S. Field technicians appreciate these experiences because they help make up for the difficult hours and circumstances that are too often a mainstay in our line of work. After taking in the breathtaking scene at the job site, we got to work setting up our equipment at the bypassed OFC.

Instruments

The devices used consisted of an advanced dual-sensor (ultrasonic and TEV) handheld PD instrument (Figure 2) for equipment condition assessment and a computer-controlled cable PD assessment instrument equipped with advanced noise removal algorithms (Figure 3) and high frequency current transformers (HFCT).

Safety

Anyone experienced in our industry understands the importance of safety, but it’s important to re-emphasize it — especially in the context of on-line testing — as these tests are performed in a different manner and not as frequently as offline tests. All of the general safety practices apply to on-line testing, but additional rules must be followed. Detailed safety procedures will be presented in a future NETA World article, but are briefly summarized below:

FIELD SURVEY

Before thinking about testing cables, consider if it’s electrically safe enough to remove the cover and begin the testing sequence. First, perform partial discharge detection of the bypassed OFC using the handheld dual-sensor instrument. If partial discharge activity is present, it’s best to not open the enclosure at all. Alternatively, we may choose to very carefully remove the cover and perform visual inspections. However, it is very unlikely that we would proceed with performing cable PD testing if the handheld instrument detected a problem.

In this case, the handheld instrument test did not detect any PD activity, so the OFC cover was carefully removed and the internal components were visually inspected. We determined that the split-core high-frequency sensors could be safely installed at the new splices (Figure 4).

We then recorded the signals using the computer-based system. We monitored all three cable phases simultaneously and used a magnetic TEV sensor on the enclosure to monitor local discharge activity. A standard 120v outlet was readily available to obtain our synch signal, which is critical for determining precise phase resolved patterns. PD occurs as the voltage increases near maximum positive and negative levels of the sine wave, and the synch signal allows confirmation that true PD is present and producing a strong, repeatable pattern across a full cycle.

After the recording was complete, we analyzed the test results. C-phase displayed the nearly synchronous pattern shown in Figure 5, indicating possible PD. Next, we analyzed the waveforms to see if they also displayed PD characteristics. It was first noted that the waveforms had fairly long pulse widths, which could be indicative of distant PD or noise.

Further waveform analysis (Figure 6) determined that the signals were noise. The instrument’s automatic noise-removal algorithms are excellent, but it is still necessary to manually analyze the waveforms further to confirm PD. The OFC cover was reinstalled, and we went on to the next location.

The distribution system consisted of several OFCs throughout the facility connected in series (line-side daisy-chained) with the load side feeding dry-type transformers, so we set up at the next OFC. None of the downstream OFCs would permit safe access for HFCT installation. With three sets of cables in the small enclosure, years of dust buildup, and several oil leaks apparent, testing was unnecessary — the terminations would all need repair, and the aged OFCs presented a major hazard themselves.

We used the handheld instrument to determine if any OFC or dry-type transformer failures were impending. All dry-type transformers were discharge-free, and one major problem was pinpointed with the ultrasonic sensor (Figure 7) in a short, unshielded jumper cable segment connecting two OFCs (Figure 8).

Assessment Results

The final assessment determined that the incoming circuit was free of PD activity and a short OFC-to-OFC jumper circuit needed immediate replacement. Additionally, it was recommended that all OFCs be replaced with air- or gas-insulated switches, and repairs to all associated terminations conducted simultaneously.

Conclusion

Following a major termination failure, the client was uncertain of the incoming cable integrity as well as the condition of their other medium voltage equipment. Although upcoming results of off-line cable tests may prove otherwise, it is expected that this cable segment is good and replacement can be deferred. This case study is an excellent example of how on-line PD testing helped a customer determine the best use of their funds, resulting in a much wiser replacement of the failing jumper segment and dangerous OFCs instead of the previously recommended replacement of the entire MV cable system.