Rajarshi Roychowdhury, Xuan Wu, Michael Russ, Dennis Fleming, Joshua Spalding
Transformers have been used in the power industry for more than a century since the widespread use of ac generation, transmission, and distribution became the standard. Even now, they remain one of the most important substation assets that utilities around the world plan and operate. Figure 1 shows a 200-MVA transformer in service at a utility substation. In the past few decades, with a rapid push toward grid digitalization due to the proliferation of distributed energy resources (DER), the age-old transformer asset management strategy has also evolved for the better.
Modern power transformers are often designed for higher flux densities to reduce the transformer size. However, with higher flux densities, the core loss and resulting increased transformer temperature can become a problem. With the increased penetration of DERs, reverse power flow also becomes challenging to handle unless utilities pay attention and specify new transformers accordingly. Utilities around the globe should plan for effective predictive maintenance strategies, as falling back on “overdesign” might not be relevant anymore.
Utilities must strive for a holistic transformer asset management strategy that clearly defines the structure being followed based on periodic standard data collection, monitoring packages, and tests. The following sections will be a quick tour of transformer loading capability calculations, impacts of reverse power flow on transformers due to DER growth, economic analysis of transformers, possible spare transformer strategies in light of enhanced data and supply chain challenges, and reliability implications of transformer failures.
The transformer loading capability calculation is a significant element in determining a transformer’s normal short-term emergency (STE) and long-term emergency (LTE) mega volt-ampere (MVA) ratings. A normal rating is generally the 24-h continuous rating. However, rating practices vary among different countries and among organizations within the same country. Sometimes, ambient temperatures are considered for calculating ratings. An STE is something shorter than 24 h and can be anything from 15 min to even 4 h, depending on which region of the world we are talking about. The same argument goes for an LTE, which is longer than an STE but shorter than 24 h. There are entities in the United States that consider the same MVA rating for both STEs and LTEs. Such ratings are typical inputs into transformer models used in planning studies and help determine loading capabilities in real-time operations.
There are several ways a utility can define how to rate its transformers, which range in degrees of complexity. However, it is critical that the methodology be defined and followed to maintain compliance with certain mandatory North American Electric Reliability Corporation standards, such as FAC-008, which mandates that each utility develop and maintain a rating methodology for bulk electric system transformers. Before delving more into transformer ratings that are followed by utilities, it is pertinent to have a brief discussion of the transformer core flux densities at this point to set the ground for the reverse power flow that will be discussed later.
International Electrotechnical Commission (IEC) 60076-1-2011, which is followed by many nations worldwide, specifies a voltage/frequency (V/f) of 110% of the rated V/f for the continuous no-load operation of power transformers. The flux density in the core is directly proportional to V/f if the number of turns and the core area are kept constant for a transformer in service. As per the IEC, along with the no-load requirement, power transformers should not be operated at a flux density over 5% more than they were designed for. Generally, for cold-rolled grain-oriented silicon steel, which is one of the most popular core materials, 1.9 T is the absolute maximum flux density that a transformer might operate at to have the additional 5% headroom for core saturation. Generally, and though it might differ slightly around the world, flux densities of 1.7 T for generator step-up transformers and 1.6–1.65 T for others are often used. A flux density lower than these values might also be used for specific reasons, such as low-noise requirements for transformers used in urban areas and mitigating reverse power flow impacts for areas that are predicted to witness DER growth.
Now, coming back to transformer ratings, multiple ways exist to define a rating methodology for bulk electric system transformers and other lower-voltage transformers. For instance, a utility may simply use a transformer’s nameplate rating as its normal operating rating and 110% of its nameplate rating as its STE rating, based on good utility practice and engineering judgment. Design and manufacturing margins cause the temperature rise test results at the transformer nameplate rating to be less than the defined temperature limit values. This way is simple to justify, document, and review but lacks technical details and possibly underestimates a transformer’s loading capability, which may negatively impact grid capacity.
A more rigorous way of determining a transformer’s loading capability is using IEEE C57.91 clause 7 equations, which are used to calculate the operating temperatures (oil and winding hottest spot) of a transformer, based on the load cycle and ambient temperature. The equations can also be used to determine the reduction of a transformer’s insulation life referenced to a benchmark value. The allowable loading capability for any defined load cycle and ambient temperature profile is determined by iterating the calculations until any one of the temperature, loss of life, or maximum percentage of nameplate rating limits is reached.
A simplified calculation process for clause 7 is described in Figure 2. Measured temperature (i.e., ambient, winding hottest spot, winding average, top oil, and so on) and kilowatt loss data (i.e., load and no-load losses) are obtained from factory tests based on rated loads. As a result of considering the site-specific ambient temperature profile and the iterative process to identify the maximum allowable load (mega volt-ampere) rating, the ultimate load is likely greater than the rated one, which can be used as a new loading capability rating and provide more capacity headroom for a dispatcher. In summary, clause 7 equations are needed to derive the ultimate transformer operating temperatures with ultimate loads and ambient temperatures. The calculations are included in an iterative process, which gradually increases a transformer’s ultimate load until one of the criteria is reached (typically the winding hottest-spot temperature threshold). Such an ultimate load will be documented as the transformer’s loading capability or MVA rating associated with the input load and ambient temperature profiles.
To the best of our knowledge, some utilities adopt the clause 7 methodology by assuming a constant load (e.g., one per unit) and average daily ambient temperature (e.g., 30 °C). This methodology does not consider load and ambient temperature variation in the calculations, which simplifies the process but leads to a potentially more conservative loading capability result. As detailed in Figure 3(a), the ambient temperature (yellow bars) and the transformer loading (black line) are assumed constant. When the transformer is loaded to 1.15 per unit (the black line), the calculated winding hottest-spot temperature reaches its limit (120 °C), as shown by the red dashed line. In other words, the transformer’s loading capability is 1.15 per unit, with a binding constraint at the winding hottest-spot temperature.
As shown in Figure 3(b), if a variable ambient temperature (yellow bars) and transformer load cycle curve (black line) are considered, the resulting winding hottest-spot temperature (red dashed line) and top oil temperature (blue dashed line) curves are obtained. In this example, the winding hottest-spot temperature reaches its limit (120 °C) at 9 p.m., which is the binding constraint limiting the transformer loading capability at 1.23 per unit when the peak load occurs at 7 p.m. Note that the load cycle curve [such as the black line in Figure 3(b)] is defined with its constant shape but can be shifted up or down until a constraint (the winding hottest-spot temperature or top oil temperature) is binding. This comparison shows that considering variable ambient temperature and load cycle curves likely increases a transformer’s loading capability because a transformer is loaded with variable loads that help dissipate heat.
We also benchmarked how variable load cycle curves impact a transformer’s operating temperatures. In Figure 4, a load curve with a solid green line represents a typical “duck” load shape, which involves a lower demand when sufficient solar radiation is available around 2 p.m., followed by a rapidly increasing demand ramp from 3 to 6 p.m. To the contrary, the other load curve with a dashed green line pertains to a traditional load shape for a region with an insignificant amount of DERs. Note that the DERs considered in this article are mainly solar rooftop systems. However, other forms of DERs, including battery energy storage systems, should be evaluated for transformer loading capability calculations and their potential impacts. The red solid- and dashed-line curves represent the calculated winding hottest-spot temperature results pertaining to the “duck” and traditional load shapes, respectively, which are both limited at 120 °C as desired. The peak loading capability of the transformer for the “duck” load shape is higher than that for the traditional load shape. A transformer energized under a “duck” curve can bear a higher peak load, as the heat inside the transformer is accumulated starting from a lower temperature (at 2 p.m.). Also note that the trend of a transformer’s operating temperature (i.e., the winding hottest-spot temperature) always lags that of the load curve. For example, the peak of the “duck” load curve (solid green line) occurs at 6 p.m., while the peak of its corresponding winding hottest-spot temperature (solid red line) occurs two hours later, at 8 p.m. This delay between the transformer peak load and peak winding hottest-spot temperature is due to the time constant for a transformer’s oil temperature change. In other words, a transformer’s operating temperatures cannot change immediately with the load change, as the transformer oil has a high heat capacity.
Based on IEEE-C57.91, during transformer overloads for STE and LTE operations, the temperature of the oil in the winding cooling ducts rises rapidly at a time constant much shorter than the bulk oil inside the tank. Therefore, during this transient condition, the duct oil temperature adjacent to the winding hottest-spot location is higher than the top oil temperature in the tank. Figure 4 illustrates this occurrence, where the orange curve representing the winding duct oil temperature surpasses the blue curve representing the tank top oil temperature right after 3 p.m. Note that, at 3 p.m., the load ramps up due to the decrease of photovoltaic generation. This situation occurs because the winding duct oil reacts to the ramp of the load more quickly than the reaction of the tank oil, due to a shorter time constant. This phenomenon needs to be noted because it can result in winding hottest-spot temperatures greater than calculated by the equations of clause 7, which do not consider the winding duct oil temperatures as a medium to calculate the winding hottest-spot temperatures (as shown in Figure 5). Therefore, the IEEE-C57.91 working group provides an alternate methodology to better reflect the thermal transformation among windings, duct oil, and tank oil while considering various cooling methods.
Another emerging trend surfacing in the utility space is the impact of reverse power from DERs back to the grid through transformers. Due to the large-scale penetration of DERs, there has been a growing concern about reverse power flow and its impact on interface transformers. Reverse power flow happens when power starts to flow from the distribution to the transmission system through interface transformers, due to large-scale renewable penetration producing more power than a local load requires and due to the absence of adequate energy storage capabilities in the distribution DER nodes to which the DERs are connected.
The question is, Does reverse power flow affect transformer health, and how can utilities prepare for this phenomenon in the coming years as more DERs get interconnected to the power system? Let’s consider a two-winding transformer with its high side connected to the grid and its low side connected to a load or DER. The high-side voltage is v1, and the low-side terminal voltage is v2, with v1 being the reference for comparison. Now, if the low side is connected to an inductive load, such as a motor, that draws a lagging reactive power, both active power P and reactive power Q flow from the grid side to the load side. The phase angle between v1 and v2 is greater than zero, or v1 leads v2, and the magnitude of the excitation voltage E, in this case, is greater than the magnitude of the load terminal voltage v2. This is the case where the transformer is operating in the first quadrant (+P, +Q), and the magnetizing current Im is drawn from the grid.
In the second scenario, we consider a second quadrant operation where the load side delivers active power (–P) but absorbs reactive power (+Q). In this case, active power flows from the load side to the grid, with the phase angle between v1 and v2 now less than zero, or v2 leading v1, and the magnitude of the excitation voltage E is greater than the magnitude of the load side terminal voltage v2. The magnetizing current Im is again drawn from the grid in this case. In the third scenario, let’s consider the situation when the load side delivers both active and reactive power to the grid (–P, –Q), an example of a third quadrant operation. In this case, the active power P flows from the load (DER) to the grid, while the phase angle between v1 and v2 is less than zero, or v2 leading v1, and the magnitude of the excitation voltage E is now less than the magnitude of the load side terminal voltage v2. The magnetizing current Im, in this case, is drawn from the load (DER) side. An example of when this can happen is solar photovoltaics with capacitor banks for load balancing.
In the final scenario, we have a capacitive load that is connected to the transformer load side. This can be for boosting the load voltage and to compensate for line drops in the low side. In this case, the transformer operates in the fourth quadrant, with active power (+P) flowing from the grid side to the load side and reactive power (–Q) flowing from the load side to the grid side. Since the active power flows from the grid side to the load side, the phase angle between v1 and v2 is positive, or v1 leading v2. The magnitude of the excitation voltage E is less than the magnitude of the load side terminal voltage v2, and hence, the magnetizing current Im is supplied by the DER in this scenario. In a transformer core, the magnetizing flux is set up by the excitation voltage E through the magnetizing current Im. When the transformer is operating in the first and the second quadrants, the core flux is established by the grid voltage. When the transformer is operating in the third and the fourth quadrants, the excitation voltage and, in turn, the core magnetizing flux are initiated by the DER instead. Transformers are designed such that the flux in the core is established by the grid voltage, which is tightly regulated in a certain band and for all practical purposes can be taken as constant. When the core flux is instead set up by the DER, it is seen through practical utility experience that the flux will increase in the core during reverse power flow.
The point to understand here is that transformers are optimally designed to operate close to the knee point of their magnetizing characteristics. Therefore, any increase in the core flux density will increase the transformer core losses significantly. Increased core loss might affect the thermal life of a transformer. Moreover, grid voltage being constant, if the excitation current increases due to the DER, there is a good chance of core saturation. Core saturation in a transformer creates additional losses and harmonics. A typical three-limb core-type transformer does not provide a path for zero sequence flux, and hence, increased harmonics and core saturation can be greatly aggravated in these scenarios.
Therefore, a good possibility exists that the life of a transformer might be shortened due to reverse power flow, and hence, some utilities in the United States already have specific reverse power flow requirements limiting the amount of reverse power flow through interface transformers. Also, utilities might be able to get ahead of this problem by initiating a conversation with transformer manufacturers, most of which, by the authors’ experience, are aware of this problem and currently working diligently to solve this issue in multiple ways. Increased taps for on-load tap changers and core designs with reduced flux densities are some of the possible resolutions of this problem. However, there are several legacy transformers in the system today that face the challenges outlined in this section.
For new transformers, it is important to understand these design considerations if a transformer will be installed at a place where there is high DER growth or DER growth is predicted to rise in the future. For existing transformers in places with high DER penetration, monitoring transformer harmonics at close time intervals throughout the year, depending on the geography and DER penetration, should be prioritized. Moreover, transformer primary and secondary voltage checks should provide a way to understand whether there is an issue in case reverse power flow is happening at the interface. When ordering new interface transformers in areas of the grid that are predicted to witness high DER penetration, a slightly low flux ratio can be chosen (1.5 or 1.6 T) instead of the more standard 1.7 T. Distribution planning in a high-DER penetration scenario becomes quite different and more challenging than traditional utility distribution planning, and a minimum loading criterion might need to be developed with thorough studies and expertise to prevent reverse power flow if operating with legacy transformer designs.
In addition to the challenges of planning for increased levels of DERs on circuits, the availability of transformers is another issue facing the utility industry today. Due to COVID-19 and supply chain disruptions, the power industry is now witnessing a challenge in procuring some key assets, and transformers have become particularly challenging to procure in a normal time frame. In the United States, approximately 90% of the consumed power flows through a large power transformer (LPT). Due to the shortage of domestic manufacturers, most of the LPTs are imported, and this creates a potential supply bottleneck. Utilities are now challenged with evolving their spare equipment strategies to incorporate the longer lead times for LPTs to maintain grid resilience. Moreover, LPT transportation and overall logistics are also challenging and should be planned efficiently. Figure 6 shows an LPT being transported to be put in service in the AES Ohio service territory.
In June 2022, the United States mandated the use of the Defense Production Act to increase the output of transformers for domestic manufacturers. Figure 7 presents a representative process to acquire an LPT by a utility. This process, as of writing this article, can take anywhere from 12 to 24 months, and hence, proactive planning becomes key. Figure 8 gives the global transformer trade average export value by power rating capability, and we can see that LPTs are globally the most sought-after component in the transformer manufacturing industry. Due to challenges in the supply chain owing to COVID-19 and other factors, it is essential that utilities take a proactive role in identifying transformers that are essential to manage the bulk power system reliability. It is also important for utilities to develop their own scoring model based on tests and monitoring/instrumentation data and create a sound spare transformer strategy.
As a result of the supply chain challenges facing the industry, utilities must evolve their spare transformer asset strategies to reflect longer lead times for replacement planning. Historically, utilities typically run transformers to failure and carry larger spare inventory levels to account for unplanned failures, or they elect to do planned replacements and carry leaner inventories. Utilities running transformers to failure and those that use planned replacements must update their strategies and models to account for longer lead times. Not adjusting the spare transformer strategy based on current lead times can be a problematic long-term approach, due to long lead times of 18–24 months for purchasing and receiving a new transformer. The average life expectancy of a transformer is typically 40–50 years. If a certain percentage of transformers operating in the utility footprint are nearing typical life expectancy, additional monitoring and asset replacement planning should be prioritized. This will ensure that the inventory strategy is still appropriate, given the health of the transformer fleet in this long-lead-time environment.
The goal of this section is to recommend transformer asset health monitoring strategies to transition the industry to a future state where predictive maintenance takes over preemptive maintenance due to failure. To facilitate this, utilities should develop a transformer instrumentation standard (such as the information listed in Table 1) that will ensure that data are brought back from operations and into asset management platforms used for continuous monitoring in operations and long-range strategic asset planning. This developed standard will serve as the basis for future purchasing specifications outlining the instrumentation needs for the concerned utility high-voltage power transformers. This will also serve to strategically recommend targeted instrumentation projects that address any major gaps with legacy transformers currently in service. This developed instrumentation standard document should set forth a minimum requirement of monitoring equipment to monitor asset condition.
Table 1. An example of a standardized instrumentation package for substation transformers.
Many utilities perform annual oil testing, dissolved gas analysis (DGA), and a power factor test every three to five years. These tests are used to monitor the health of a transformer and help predict when the transformer will need to be replaced. DGA is often the first indication that problems exist with a transformer. DGA consists of sending oil samples to a laboratory for testing, as transformer oil health is a good indicator of the overall transformer itself. DGA reports individual and total combustible gas generation rates based on IEC 60599 and/or IEEE C57-104 standards.
There are some recommended industry rules to interpret a DGA report. However, transformers are complex pieces of critical electrical equipment, and hence, experience and expert engineering judgment should always be consulted and are recommended. A good utility practice is to perform ™ DGA, maintaining a good bookkeeping practice of the results and comparing earlier DGA reports with later ones. However, most indicators in these tests advance rapidly, and hence, periodic testing in close intervals and comparing the results with previous test reports are some of the key aspects to be considered for DGA. Also, new engineers entering the testing field should note that if a transformer is cooled and de-energized or is a new transformer that has not yet completed at least two to four weeks of continuous service, DGA results might be quite unreliable.
Table 2 condenses some of the key recommendations and action items from a DGA analysis. If the gas concentration, in parts per million, exceeds the normal limits but is less than the action limits, the frequency of DGA testing should be increased with some consideration given to planned outages in the future for further evaluation. As the gas concentration exceeds the “action” column, removal of the transformer might be considered. It should be noted that these gas concentrations should be tested periodically, and any sharp increase in the concentrations might indicate a potential problem. In summary, just the concentration levels of these key gases should not be taken as an indication of a possible problem, but the trend of these dissolved gas levels should be evaluated to better understand DGA results.
Table 2. DGA information based on IEEE C57.104 and engineering experience.
One of the common causes of a transformer failure is the breakdown of the dielectric. Generally, dielectric breakdown happens due to repeated thermal cycles with high temperatures combined with moisture, oxygen, and other air particulates. It has been observed that dielectric breakdown in a transformer is aggravated if the transformer is overloaded frequently. A power factor test of a transformer is done to test the ac characteristics of the transformer dielectric. Dielectric losses of a transformer’s insulation can be obtained from a power factor test. When the transformer insulation is excited by a known ac voltage, it draws charging current. The charging current can be split into two components: capacitive and resistive current. As the name suggests, capacitive current leads the applied voltage by 90º, whereas the resistive current component is in phase with the applied voltage. Capacitive current is directly proportional to the dielectric constant of the insulation, area, and voltage and inversely proportional to the thickness of the insulation under test. Changes in capacitive current indicate insulation degradation. As the insulation deteriorates, more current will leak through the insulation, and the power factor therefore becomes increasingly greater. The results of a power factor test will confirm the condition of the insulation in the windings, bushings, tap changers, and oil. For modern oil-filled power transformers, power factors of 0.5% or less, at 20 °C (68 °F), for individual windings to ground and interwinding insulations are generally considered a passing criterion.
Companies employ different strategies with respect to transformer replacement. These range from run to failure to targeted replacement.
Strategy 1: Run Transformers Until Failure and Have Large Spare Inventory to Account for Failures
Some companies choose to run their transformers until there is a failure. This strategy ensures that the company has used the equipment to its fullest extent but could come at the potential expense of operational risk posed by an unplanned failure. This philosophy would typically require a company to maintain more spares so that it would be prepared to replace a transformer after it fails and be able to have sufficient spares available for multiple individual failures in a short time frame. A failure could happen at any moment and could create other issues in the system, so this justifies the need for a larger inventory of spares. It is recommended that utilities have at least a basic prediction model, if detailed predictive models are not technically feasible, combined with inventory lead time models to recommend the appropriate transformer strategy strategically and dynamically.
Strategy 2: Replacement Based on Asset Management Practices and Optimized Inventory
Most companies employ a programmatic or targeted replacement strategy for transformers, based on predictive health measures from their asset management department. Some companies replace a set number of transformers every year, and others replace based on the health of a transformer. The latter would require a formal process to assess the condition of a transformer. Predictive transformer replacements can be driven by models derived from a utility’s historical datasets, analyzing past failures, historic test results, and maintenance records to train an advanced model capable of predicting future failures. Further, advanced analytical techniques could build off this predictive model and pair it with a basic optimization model to establish optimal inventory levels based on predicted failures and dynamic transformer lead times updated regularly based on updated lead times from utility supply chain experts.
A long-term spare transformer strategy should be transitioned to a risk-based methodology including additional asset monitoring data. This strategy will help ensure that inventory levels directly align with the probability of failures to ensure an optimal spare transformer strategy. As more data become available through asset monitoring, additional performance metrics should be built into the risk formula. It is recommended that transformer inventory levels be evaluated on an annual basis to ensure that the inventory strategy is refreshed based on the risk profile.
Once a utility reaches a future state in online transformer monitoring, it may be possible to shift to planned transformer replacements, versus the current methodology of running to failure, if sufficient leading indicators are developed that can positively identify a failing transformer. This strategy would require fewer spare transformers since the methodology determines which transformers to replace and when to replace them. As a part of the change to this strategy, utilities would need to create a transformer health index, criticality index, or risk index or some combination of the three that considers everything from transformer test results, condition, and utilization to location and criticality. As noted in this article, targeted replacement is the strategy employed by most utilities, and a combination of these three metrics would help to better plan and prepare for the future. This three-pronged scoring model might be used to rank the worst-performing transformers in a utility footprint.
Health Index
Some of the recommended parameters that might be used to create the health index are DGA, load history, global loss factor, infrared thermography, oil quality analysis, Omicron testing data, Furan’s content, leakage reactance, winding resistance, core to ground connection, conditions of bushings, tank corrosion, and so on. Each of these parameters can be assigned a weight, and an overall quantitative health index can be formulated. These metrics were determined by looking at the failure modes of a transformer and understanding the maintenance and tests taken to help understand how that transformer is performing. The biggest challenge that some utilities might run into is the lack of data and the lack of consistency in the data. With the data that might be available, utilities might be able to rank all these metrics from zero to 10, with 10 being perfect health. For example, if a transformer has high gas levels, then that can be assigned a certain score threshold, depending on the severity and internal utility practice.
A lesson that might help the industry, from the authors’ collective experience, is the fact that age has a lower effect on transformer health than most people think. Transformers are created to withstand a lot of capacity. Unless there are outside forces causing a unit to fail, the age of a transformer is almost irrelevant to the overall health. That is why age can have a small weighting factor in the metrics developed. There are quite a few 80-year-old transformers in some utility territories that are still performing quite well. With all these data, metrics, and weighting factors, a model should be built to monitor the health of transformers. Transformer monitoring/instrumentation packages should be used and are recommended to augment the developed models. In the future, using these developed models, utilities should be able to develop a structure feeding data into health score models to do predictive and preventative maintenance and reduce outages. This will help utilities to shift from the strategy of running a transformer to failure. Understanding the difference between time-based maintenance and preventative/proactive maintenance is the key here. Obtaining data from the installed transformers through using instrumentation and automation would help the models make better decisions.
The authors recommend that the developed health index should be reflective of the overall transformer health and should provide an objective evaluation instead of being a subjective observation. The index should be understandable and be interpreted easily for it to be widely useful.
Criticality Index
Since the health index depends on testing data, a criticality index might also be defined for the benefit of the overall asset management plan. The criticality index will assign a score to each transformer, based on network location, load serving criticality, impact on system stability, consequence of failure, and social and environmental impacts, if any. Similar to the health index, the criticality index parameters might be weighted depending on their relative importance. For example, more weight might be given to system instability and load serving capability than some other parameters, though this should depend on engineering judgment and is best left to utilities to find out what works best for their footprint.
Risk Index
A risk index might be formulated, which is a combination of the health index and criticality index, with equal weight given to both. Since the risk index takes into consideration both the health index and the criticality index, this provides a better way for utilities to develop a holistic transformer asset management strategy.
Transformers remain one of the key substation assets that utilities around the world operate. As we move toward a more digital future, utility asset management strategies should be more proactive in taking care of this very important equipment that directly affects the reliability of the bulk power system. This article delineated some past and state-of-the-art transformer failure prediction methodologies that are used in the industry: transformer asset health monitoring concepts, instrumentation packages that are currently available, loading calculations, and the effect of reverse power flow in transformers.
Proactive transformer health monitoring using standard tests and bringing data from the field in regular intervals to feed into a health model will be key in moving to the future. Also, while planning for transformers in an area with high renewable growth, care must be taken regarding the choice of transformers and, if required, through adequate system studies. Specifications regarding low-core-flux designs might be incorporated at an early stage in these areas. The authors’ hope is that this article will inspire utilities to set up proactive standards and requirements as we transition to a smarter future with more active loads, renewables, and two-way power flow. As a final note, the authors would like to point out that the article is devoted primarily to substation transformers and might not be applicable for other transmission and distribution assets.
IEEE Guide for Loading Mineral-Oil-Immersed Transformers and Step-Voltage Regulators, IEEE Standard C57.91, 2011.
IEEE Guide for the Interpretation of Gases Generated in Mineral Oil-Immersed Transformers, IEEE Standard C57.104, 2019.
IEEE Standard for Interconnection and Interoperability of Distributed Energy Resources with Associated Electric Power Systems Interfaces, IEEE Standard 1547-2018.
Mineral Oil-filled Electrical Equipment in Service—Guidance on the Interpretation of Dissolved and Free Gases Analysis, IEC Standard 60599, 2022.
Office of Technology Evaluation, “The effect of imports of transformers and transformer components on the national security,” Bureau of Industry and Security, U.S. Department of Commerce, Washington, DC, USA, 2020. Accessed: Dec. 21, 2022. [Online.] Available: https://www.bis.doc.gov/index.php/documents/section-232-investigations/2790-redacted-goes-report-20210723-ab-redacted/file
Rajarshi Roychowdhury is with AES Ohio, Dayton, OH 45432 USA.
Xuan Wu is with AES Indiana, Indianapolis, IN 46221 USA.
Michael Russ is with AES Indiana, Indianapolis, IN 46221 USA, and AES Ohio, Dayton, OH 45432 USA.
Dennis Fleming is with AES Ohio, Dayton, OH 45432 USA.
Joshua Spalding is with AES Indiana, Indianapolis, IN 46221 USA.
Digital Object Identifier 10.1109/MPE.2022.3230889