Transformers limit 38% of all circuits in the England and Wales electrical transmission network, and ultimately limit transmission boundary power flows. Where there is a need to remove network constraints, investment in increased capacity is necessary, this project seeks to investigate whether increased capacity at the same time as allowing an increased rate of transformer ageing could be justified economically.

Transformers are rated using a tool called TRALC3, which has been developed within NGET over the course of many years (original work was conducted by the Central Electricity Generating Board). The nameplate rating is determined on the basis of the maximum load which could be applied, under certain environmental conditions, to achieve the desired life expectancy of the transformer. This means solving a thermal model, on the basis that the operating temperature of the insulation is one of the primary factors which determine the rate of ageing of a transformer. For the majority of the time, the load on the transformer is less than 75% of its nameplate rating. This means that the insulation system within the transformer ages more slowly than its standard life expectancy would suggest. However, under emergency scenarios they can carry up to 150% of their maximum rating.

National Grid is interested in investigating methods of ‘uprating’ transformers at specific points in the network where investment is likely to replace the assets in the short to medium term, allowing constraints to be removed earlier. This creates a more competitive electricity market and would allow National Grid to meet a key challenge, which is getting new customers (i.e. generators or major loads) connected sooner and/or with less ponderous investments in the expansion of the infrastructure. This project seeks to investigate how this could be achieved, with a focus on quantifying the potential economic benefits that could be realised using such an approach. This has the potential to lead to novel commercial arrangements in the future, once the underlying technical and economic drivers can be adequately assessed.

##### Objectives

The objective of the EAT project is to determine if there are scenarios in which the use of higher transformer ratings, at the expense of the longevity of the asset, could be economically beneficial to the operation of the electricity transmission system. The project would seek to design a methodology which would allow the financial impact of different options to be assessed, with due consideration of the uncertainty surrounding actual asset utilisation. This would allow greater understanding of the economic case for providing enhanced ratings.

### Learnings

##### Outcomes

It has been confirmed that there is some scope for uprating NGET transformers within (and possibly beyond) the limits required for typical lifetime expectations. However, the higher ratings would probably not allow additional connections unless they could be used as the basis of the boundary flow capability calculations.

While it seems likely that the current limit imposed by IEC 60076-7 is too restrictive in many cases, it was found during this project that obtaining the data needed to determine what current limits are actually required would be very difficult. With sufficient transformer-design data, it might be possible to rely on finite-element modelling to predict the flux densities in the core and the levels of eddy-current heating, but obtaining sufficient data seems unlikely. With enough test data to show that (at rated current) eddy-current heating outside the windings was insignificant, it would still be necessary to check whether high loads would cause saturation of the core. Use of finite-element modelling for this would require much less data. In the situation where insufficient data would be available for either model, the current limit used in the existing transformer thermal ratings models (TRALC3) must be retained.

As the current limit restricts short-term rating enhancements any flexibility implied by potential higher boundary capacity may not be realised. A new cyclic rating could be defined (for example based on 100 hours/year at temperatures up to 110 °C with temperatures below 86 °C for the rest of the year) that would be consistent with current transformer ageing expectations, but this would probably not allow additional connections unless the higher rating could be used as the basis of the boundary flow capability calculations.

**Recommendations for further work**

The largest impact on the overall ageing of a transformer is mostly due to the load it is subjected to when a continuous rating is applied. It would thus be worth investigating the possibility for NGET to choose a set of ratings in such a way that the ageing, as determined by the power flows the NGESO would transmit over the network subject to the given ratings, is minimised. It would be advantageous to focus on the problem of choosing an optimal continuous rating for the set of transformers present in the portion of the power grid at hand in order to minimise their ageing while guaranteeing that power can be feasibly dispatched (by checking the feasibility of the optimal power-flow problem the NGESO would solve).

Some additional issues for future work have also been highlighted in the deliverables, including those related to other unaccounted sources of transformer ageing, practical implementation challenges due to automatic tap changer control and consideration of non-homogeneous sets of transformers in a circuit.

##### Lessons Learnt

The Optimal Power Flow (OPF) models and algorithms (which, in their classical version, address a steady-state scenario) have been extended to tackle dynamic situations where load and generation fluctuate over time. The results from this activity can provide the mathematical means to design a strategy to optimise flexibly different aspects of the power grid from the perspective of both NGET (for aspects of, e.g., network expansion) and NGESO (to simulate, e.g., the impact of cost constraints on the system).

From an algorithmic perspective, one key extension of this project could be the development of a technique for finding a set of ratings which minimise the transformer ageing determined by the power flows the NGESO would transmit as a consequence of the chosen ratings. Such a problem, which is of bi-level nature, could be solved by building upon the tools that have been constructed in this project. In particular, the Julia/JuMP+Matlab/TRALC2 code could be used as a subroutine to evaluate the feasibility of a certain choice of ratings as well as for computing the temperatures and, ultimately, the ageing, such a choice of ratings would determine for a set of pre-defined scenarios. A method would have to be determined for exploring the rating space, iteratively moving from a rating vector to another in such a space in order to find a (locally) optimal solution.

It seems likely that the current limit imposed by IEC 60076-7 is too restrictive in many cases. The project has revealed that a strong model of the way the NGESO would distribute power over the network subject to enhanced ratings is crucial to assess accurately how large a reduction of ageing said updated ratings could lead to.

Easier access to data (such as, for instance, historical loads or the amount of real/reactive power generated at different buses over time) would allow for the production of stronger supporting evidence for the results of many case studies than can be obtained with synthetic data.

**Dissemination**

Opportunities for dissemination were limited by COVID-19. Some elements of the work carried out were presented at the 3rd IMA/OR society conference which took place between the 20th and the 23rd of April 2021 virtually via Zoom.

A journal paper on the work is also planned to be completed after the project has completed.