Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Risk-Based Dynamic Thermal Rating in Distribution Transformers via Probabilistic Forecasting

Scott Angus, Jethro Browell, David Greenwood, Matthew Deakin

Abstract

Low voltage (LV) distribution transformers face accelerating demand growth while replacement lead times and costs continue to rise, making improved utilisation of existing assets essential. Static and conservative protection devices (PDs) in distribution transformers are inflexible and limit the available headroom of the transformer. This paper presents a probabilistic framework for dynamically forecasting optimal thermal protection settings. The proposed approach directly predicts the day-ahead scale factor which maximises the dynamic thermal rating of the transformer from historical load, temperature, and metadata using clustered quantile regression models trained on 644 UK LV transformers. Probabilistic forecasting quantifies overheating risk directly through the prediction percentile, enabling risk-informed operational decisions. Results show a 10--12\% additional capacity gain compared to static settings, with hotspot temperature risk matching the selected percentile, including under realistic temperature forecast errors. These results demonstrate a practical approach for distribution network operators to take advantage of PDs with adaptive settings to maximise capacity and manage risk on operational time scales.

Risk-Based Dynamic Thermal Rating in Distribution Transformers via Probabilistic Forecasting

Abstract

Low voltage (LV) distribution transformers face accelerating demand growth while replacement lead times and costs continue to rise, making improved utilisation of existing assets essential. Static and conservative protection devices (PDs) in distribution transformers are inflexible and limit the available headroom of the transformer. This paper presents a probabilistic framework for dynamically forecasting optimal thermal protection settings. The proposed approach directly predicts the day-ahead scale factor which maximises the dynamic thermal rating of the transformer from historical load, temperature, and metadata using clustered quantile regression models trained on 644 UK LV transformers. Probabilistic forecasting quantifies overheating risk directly through the prediction percentile, enabling risk-informed operational decisions. Results show a 10--12\% additional capacity gain compared to static settings, with hotspot temperature risk matching the selected percentile, including under realistic temperature forecast errors. These results demonstrate a practical approach for distribution network operators to take advantage of PDs with adaptive settings to maximise capacity and manage risk on operational time scales.
Paper Structure (22 sections, 6 equations, 8 figures, 1 table)

This paper contains 22 sections, 6 equations, 8 figures, 1 table.

Table of Contents

  1. Introduction
  2. Automated Dynamic Thermal Rating
  3. Thermal Modelling
  4. Basic Thermal Model
  5. Accounting for Unbalanced Phase Loading
  6. Protection Modelling
  7. Dual Time Constant Model
  8. Retrospective Scale Factor Calculation
  9. Probabilistic Forecasting Methodology
  10. Clustering Approaches
  11. Training
  12. Feature engineering
  13. Single- and Multi-Temperature Training
  14. Probabilistic Forecasting
  15. Quantifying Probabilistic Forecast Performance
  16. ...and 7 more sections

Figures (8)

  • Figure 1: The proposed adaptive transformer protection scheme considered to enable automated DTR. By choosing the PD's scale factor, $k$, the current at which measured phase currents $I_A,\,I_B,\,I_C$ trip can be adjusted, accounting for the impact of variable ambient temperature and off-peak load conditions on winding hotspot temperatures.
  • Figure 2: Flowcharts summarising the direct prediction of scale factors (a) and the calculation from load predictions (b).
  • Figure 3: Clustering results using PCA weighted scaled features. Axes show the magnitude of the first two principal components.
  • Figure 4: Coverage of true scale factor by predicted percentiles per transformer ($n=644$) for different training and test temperature scenarios. ST and MT represent Single Temperature and Multiple Temperature training respectively, while CP and NP represent Clean and Noisy temperature Predictions respectively. The vertical axis gives the percentage of transformers where coverage of the 90% prediction interval is less than Coverage. An inset shows the similarity of the coverage distribution for the single temperature training approach under clean and noisy temperature forecasts as well as the multiple temperature training approach under clean forecasts.
  • Figure 5: Scale factor $k$ probabilistic predictions for the transformers with the best (a) and worst (b) coverage (across the full sample of $n=644$ transformers) during the holdout period.
  • ...and 3 more figures