Conference Agenda

During routine on-line partial discharge (PD) spot measurements on high-voltage cables in a European refinery, unexpected discharge activity was detected originating from an adjacent 110 kV/6 kV, 63 MVA power transformer. Subsequent IEC 60270-compliant electrical PD tests confirmed elevated discharge levels at the high-voltage neutral, varying with the on-load tap changer (OLTC) position. Acoustic measurements localized the PD sources near the OLTC or regulating winding, suggesting potential insulation degradation in those regions.

To enhance both safety and diagnostic efficiency, the transformer was retrofitted with permanently installed PD coupling units, reducing future measurement durations from up to two days to approximately twenty minutes. A Megger ICMmonitor portable system was subsequently utilized for temporary on-line PD monitoring, which successfully captured sporadic void-type PD activity not observed during short-term spot tests.

Correlation with dissolved gas analysis (DGA) provided complementary diagnostic insights. Although elevated hydrogen levels were detected, their pattern indicated low-temperature contact resistance issues rather than PD-related in-oil degradation. The integration of PD monitoring and DGA thus enabled a more comprehensive understanding of transformer condition, distinguishing between electrical discharge phenomena and thermally induced faults.

This case study highlights the effectiveness of combining on-line PD monitoring and DGA as a practical, non-intrusive strategy for transformer condition assessment. The approach enhances diagnostic reliability, supports informed maintenance decisions, and contributes to cost-efficient asset management. Furthermore, the installation of permanent PD sensors establishes the foundation for continuous monitoring, facilitating early fault detection and proactive intervention. As a future step, implementing permanent on-line PD monitoring will provide real-time insight into insulation health, helping extend the service life of critical power transformers while minimizing unplanned outages.

References

[1] D. Hering, “Keeping a watchful eye on transformers,” ET Magazine, no.2, pp. 88–93, 2023.

[2] D. Gross, M.Soeller, “On-site Transformer Partial Discharge Diagnosis”,

IEEE International Symposium on Electric Insulation, San Juan, Puerto Rico, 2012

[3] CIGRE Technical Brochure 861: Improvements to PD Measurements for Factory and Site Acceptance Tests of Power Transformers, CIGRE, Paris, France, 2022.

The high-voltage insulation system of transformers is very complex. This system has many components: paper, pressboard or high temperature insulation materials surrounding conductors immersed in a large volume of mineral oil or ester fluid, solid (pressboard or high temperature insulations) and/or liquid insulations protecting the core and metallic transformer parts, vertical pressboard or high temperature material barriers between LV and HV and adja-cent phases etc. In high voltage all insulation components are subjected to large dielectric and creep stress.

In this paper we present dielectric and creep stress calculations using 3D or 2D finite element methods to verify the critical areas having the highest stress which can be source of failure. The location of critical area and critical stresses limits calculated were validated comparing calculation values with test data. The transformers were test-ed at higher dielectric test levels using impulse (full and/or chopped waves), switching, applied voltage and in-duced voltage tests. In all dielectric tests, we also completed PD measurements. Validation of the calculations was also performed on the investigation of failures of transformers used in service without adequate maintenance.

Calculations of dielectric and creep stresses in certain critical areas in solid or liquid insulations located in the vicinity of cable winding connections or cable connections between winding outputs and transformer accessories such as bushings and tap changers will also be presented in this paper.

The transition towards renewable and more sustainable energy sources requires an adaptation of all the components of the electric grid, from generation, via transmission to end customer. One direct consequence of the transition to renewable generation is, for instance, the addition of new power generation sources placed in floating wind parks comprising fully assembled substations. The new applications need to be carefully designed with particular attention to the mechanical resistance over time of all its components, among others, transformers, which are the focus of the present investigation. The mechanical loads arising during assembly (lifting, compacting) and transportation (on land and/or at sea) and in service (e.g., effects in-rush currents, short circuit, and seismic loads) dictate the current design rules of transformers. The new applications subject transformers to a different set of loading conditions including vibrations, long-term low frequency mechanical loads and in-service temperature fluctuations. Therefore, transformers must also be designed to withstand these loading cases over their entire life in service. To tackle the new challenges, simulations and tests on subset and full transformer active parts were performed. Different approaches were used to define the suitability of the present design to the dynamic loads arising in floating platform. The mechanical endurance of a power transformer is related to the mechanical stability of its insulation components, mainly consisting in paper-based materials. These components have proven to be reliable for standard applications in transformers so far. Static tests performed on base material can ensure the quality of the components, but when it comes to dynamic long-term performance, additional tests are needed. Paper-based materials are highly orthotropic, exhibit a time-dependent behaviour and show a certain degree of temperature and moisture dependency. These factors might also depend on one another, thus increasing considerably the complexity of the mechanical description. Some adaptations of the material characterization are required as a function of the loading to which the material will be subjected in service conditions. Low-frequency and long-term loads, typical of offshore platforms, can also be included as another type of dynamic load on a transformer. The impact of the above-mentioned loading conditions on transformer solid insulation was properly assessed in ad-hoc design experiments, where the test pieces were immersed in insulation oil and dynamically loaded over extended periods. Further steps are needed to qualify the full transformer solution. FEM models, including advanced material descriptions can provide more accurate results to forecast and avoid possible issues leading to malfunction and potentially failures. An additional step to ensure the stability of transformer was taken by testing windings under dynamic conditions. Strain gauges, load cells directly applied on the test objects recorded the changes in the applied forces on the windings over the course of the dynamic load. FEM simulations were performed to verify the assumptions and models used for the calculations. The objective of this paper is to show how the combination of focused mechanical testing and advanced mechanical modelling can help manufacturers to improve design and provide a safer final product.

Dielectric Frequency Response testing, as outlined in IEEE Std C57.161-2018, is a critical non-destructive diagnostic for assessing the moisture content, conductivity, and aging of oil-paper insulation in liquid immersed power transformers. Connecting large MVA power transformers for Dielectric Frequency Response testing poses significant challenges primarily due to the scale of the equipment, induced voltages, electromagnetic interference from nearby energized lines, and the complexity of transformer insulation systems. With the intricate structure on large MVA transformers encompassing the terminal capacitance, large bushings, inaccessible terminal point to short the leads as well as multiple parallel paths just makes it harder for the test equipment to run an effective test as it would when it approximates single winding to ground network. The consequence is a distortion of the low frequency dielectric response spectrum where parasitic capacitances between bushings and tanks, or between tertiary winding and core, may mask the true low frequency slope attributed to moisture in the paper insulation. Increased length of the test lead can also add to the absolute capacitance creating a bias in low frequency zone. While shorting the bushings is a standard practice to bring all the windings on one side to an equipotential point, it is not always convenient while running the Dielectric Frequency Response test on large MVA transformers. We discuss the workaround in these situations and best possible methods to obtain the test results that we would achieve on a conventional Dielectric Frequency Response testing with the outcome of analyzing the insulation health on that transformer.

Incipient turn-to-turn insulation faults in the secondary of protection class current transformers can be difficult to find since they often don’t show up during regular testing. These incipient faults can be dangerous since they can show up only when enough voltage is developed in the secondary of the CT, which usually happens during power system fault conditions where the secondary current rises rapidly and so does the voltage developed in the secondary as a consequence. The most common insulation test performed on current transformers, a DC insulation test, won’t typically find a turn-to-turn insulation fault since the voltage is applied between the primary and secondary windings or between them and ground which does not really stress the turn-to-turn insulation. Turn-to-turn insulation failures will only show in DC insulation tests when the fault has already burned through the insulation into the CT enclosure. Other common tests performed like ratio tests could show the issue, but the turn-to-turn fault does not show up in ratio tests unless the fault has already been declared and one or more turns have already been shorted to each other which will significantly affect the ratio measurement. If the fault is incipient but has not been declared the ratio test will pass without issues since not enough voltage is developed between the turns to declare the insulation failure. A CT excitation curve test will only find a turn-to-turn insulation fault if enough voltage is developed in the secondary winding during the test to have it “break the fault”. Other common methods of determining the excitation characteristic of the current transformer will not be able to find these incipient faults since they usually rely on injecting lower voltage levels. Surge testing, a common method used to detect turn-to-turn faults in rotating machines, can be used to detect turn-to-turn faults in current transformers, but its validity needs to be verified. Additionally, a comparative method such as sweep frequency response analysis is included in the testing as part of the investigation to verify whether it can be used and in which context to detect such faults. This is a comparative method which can be used to detect such faults once they have been declared since it uses a very low voltage for the testing. In the paper the previous three methods are used to detect incipient and already declared turn-to-turn insulation failures and the results are compared.

A physics-based thermal model was incorporated into IEEE C57.91 during its 1995 revision, building on the original work by Pierce, which was published as a paper in IEEE Transactions in 1994. This model uses an energy-based approach to integrate heat generation and its distribution among the various components of a transformer. Through this method, it enables the simulation of transient temperature conditions without relying on estimated time constants. In the latest version of the loading guide, the model was moved from the informative Annex G to the main body of the document, thereby becoming the recommended method for estimating the dynamic temperature of transformers.

Although the model was originally applied to estimate the hottest spot temperature by considering only the most critical coil, its use in online monitoring of power transformers requires temperature estimation for each individual winding. The “intuitive” approach of including all windings with their respective loss values does not produce consistent results. Consistency in this context refers to a basic validation in which the temperature of a winding remains the same whether the simulation includes only the most critical winding or all windings. To achieve this consistency, further modifications to the algorithm were necessary.

Beyond the capability to simulate multiple concentric windings, the model was also extended to estimate temperatures for windings that are axially displaced. An example of this configuration is a transformer with one low voltage and two high voltage windings per phase. In such cases, the insulating liquid flows sequentially through the lower and upper coils, which requires additional adjustments to the thermal model. All these enhancements were implemented for five different insulating liquids: mineral oil, natural ester, synthetic ester, silicone, and less flammable hydrocarbons.