Instant Payback – Building Trust in Online Monitoring

Presented By:
Marcolus Sullivan, Ory Shoemaker, Timothy McNamara
Duke Energy
Kwasi Yeboah & Randy Cox
GE Grid Solutions
TechCon 2023


Examine real-world applications of online monitoring on old and new transformers, where instant gratification is achieved by detecting and alarming on issues. The utilization of near real-time data, validated by history of maintenance data, provides subject matter experts the necessary confidence required for trusting online monitoring data. Furthermore, immediate information gathered from continuous monitoring allows for faster decision-making while determining when to remove or place a transformer in service.


Duke Energy is an electric and gas utility company, headquartered in Charlotte, NC. The company’s electric business serves more than 8.2 million customers over six states. Its service territory covers approximately 91,000 square miles with 31,000 miles of transmission lines, generating capacity of 50,000MW, and more than 10,000 power transformers. This makes it one of the largest utilities in the USA. For years Duke Energy (DE) has utilized transformer online monitoring (OLM) in its various regions. In 2015, a decision was made to further invest in the benefits of OLM. As a result, DE invested in a Health and Risk Management (HRM) analytic platform that could manage big data captured from its online monitoring devices. The key benefits of this investment were to extend asset life cycle, reduce unplanned outages, prevent catastrophic failures, and improve asset reliability.

Figure 1 – Service Territories – NC, SC, FL, IN, KY, & OH

Case 1: DGA Save

This was a case where OLM revealed a problem in a seemingly normal operating asset. A transformer located in a Midwest substation was placed in service prior to 1993. The transformer was rewound in 2003 after a winding failure occurred. The rewound transformer showed no signs of concern after the repair. In 2009, the transformer started gassing. An internal inspection revealed damage to the De-Energized Tap Changer (DETC), repairs were made, and the transformer returned to service. Several years later during routine oil sampling, the transformer appeared to still have elevated gas levels – nothing of great concern – yet a request for adding online monitoring was issued as the investigation continued.

Figure 2 – Shell Type Auto Transformer

On December 17, 2020, a dissolved gas analyzer (DGA) and bushing monitor (BHM) were installed on the transformer. During final online monitor checkout, a relay technician noticed the DGA caution light was illuminated and the DGA monitor had begun sampling every four hours. Upon further evaluation, the acetylene reading had changed from 1ppm to above 4ppm. No DGA alarm setting adjustments were made since the monitor had just been placed into service. Over the next four hours the gas value had migrated to alarm status[1] where the acetylene reading had increased to more than 7ppm and the online DGA monitor changed sampling to every hour. The Transformer SME remotely monitored the gas readings from the online DGA monitor and began comparing the newly recorded data with historical maintenance data in TOA.

Figure 3 – Transformer gassing increased under load

At 12:00pm the following day, the autotransformer load was removed but windings remained energized from the 138kV source. At this point, the acetylene values had reached over 43ppm. Upon further investigation, the other Hot Metal Gases (HMG) increased as well, particularly ethylene and methane, which mimicked the acetylene increase. As a result of load being removed, the gassing stopped. The transformer load paths were switched and tested from 138/69kV and then to 138/13kV. Both load paths caused increased gassing when serving load. Due to the transformer continuously gassing, it was decided that the transformer should be removed from service for further investigation. The transformer was successfully de-energized on January 6, 2021, under a planned outage. It was electrically tested. The core ground was tested at 500V and 1000V but failed to read above 10,000 ohms. Also, a winding resistance test was performed, and no issues were found, so the core ground seemed to be the problem.

Figure 4 – Debris and damage found in transformer

An internal inspection revealed nothing obvious with the core ground leads, but damage and debris were found throughout transformer oil. Broken dowel rods dislocated winding spacers, cellulose residue with metal particulates, broken wood supports, and torn paper were found throughout transformer. The transformer had experienced enough through-faults over its years of service, that if placed back in service imminent failure would occur. It was decommissioned and replaced with a new transformer. The OLM equipment was moved to the new bank.

Figure 5 – Broken dowels and debris from transformer

Case 2: BHM Abnormal Condition Detection

In this case, a new transformer quickly developed a problem after about two months of service. In November of 2021, two 230/100kV autotransformers, bank 9 and bank 10, were energized in a Carolina’s substation, both equipped with online dissolved gas analyzers and bushing monitors. Each installation and energization went well. On January 16, 2022, the Carolinas experienced its first snow of the year that lasted several days. On the 26th, bank 10 bushing monitor detected a change in the secondary C-phase bushing power factor on the polar plot chart. The capacitance percentage began dropping while the power factor percentage increased, yet both readings stabilized resulting in a polar plot alarm on the bushing monitor. The C-phase leakage current decreased slightly but stabilized also. A decrease in capacitance, %C1 change, indicated that this issue was something other than bushing degradation. In a failing bushing, one would normally expect an increase of leakage current which translates to an increase of capacitance change. The initial thoughts were that the bushing sensor adapter possibly could have moisture ingress. Plans were made to take an outage and replace the bushing sensor adapter.

Figure 6 – Bushing tap adapter and fill-plug tape wear damage
Figure 7 – Power Factor percentage changes of C-phase

The outage for bank 10 was scheduled for March 8th. Upon inspection of the C-phase bushing sensor adapter, the internal tap cavity showed signs of water ingress. It was suspected at the time of inspection that the bushing fill port had allowed water ingress into the tap cavity via the Teflon taped fill plug. Several photos were taken of the issue. The bushing sensor adapter was removed for further analysis and replaced with new sensor adapter. The transformer was re-energized without connecting the secondary C-phase sensor adapter due to unclear guidance on the responsibility of the fill-plug repair. Meanwhile, the bushing monitor vendor performed a root cause analysis on the original bushing sensor tap adapter. It was examined for water ingress. No signs of ingress were found. The sensor adapter passed all pressure and electrical tests. Furthermore, it was confirmed by photos that there was water ingress into the bushing tap cavity via the fill plug. Another outage date was scheduled for May 18th to reinstall the bushing adapter. After all repair work was completed, the secondary C-phase online monitoring returned to its normal operating range for leakage current, capacitance and power factor. In light of this event, an inspection of the test tap fill ports is now performed as a matter of procedure. This finding has allowed for the detection of similar issues in subsequent online monitoring installations.

Figure 8 – Capacitance and Power Factor percentage plots


In summary, the use of online monitors provided immediate real-time information for a more informed decision-making process on asset health and operation status. The Transformer SME gained confidence in the online monitor data to avert an imminent transformer failure in Case 1 and to prevent bushing damage in Case 2. These two cases demonstrated the benefits sought in Duke Energy’s decision to invest in online monitoring and to believe in having a robust HRM analytic platform that enables decision-making based on real-time information in an evolving environment.


[1] IEEE Standard C57.104-2008

Join our email list

We use cookies to give you the best online experience. By using this website you agree with our cookie policy.