Learnings

Outcomes

FEA Models

The FEA models allow for rapid calculation of temperature and ratings for complex cable installations which are not always amenable to the IEC standards. The FEA models have also been used to challenge the IEC standards, demonstrating issues with the external thermal resistance for large bipole systems which can lead to errors in temperature of ~1°C at high loads. It has also been demonstrated that FEA models combined with geotechnical information can reveal spurious DTS temperatures. 

Visualisation Software

The software developed to visualise DTS datasets allows for interactive analysis of ≥100 GB DTS archives from HVDC cables and separates load-related structure from baseline drift using statistical analysis. Analysis of the DTS data can provide significant additional insights into the thermal properties in both time and space and, when combined with FEA models, can help optimise cable designs. Being able to rapidly analyse large volumes of DTS data also enhances our ability to monitor cable performance over time and identify any changes in the ground condition (e.g. decreased sediment cover) that could lead to exposure and increased failure threat to the cable.  

Machine Learning

The modified temporal fusion machine learning architecture developed as part of AnTICS achieves a median root mean square error (RMSE) of approximately 0.5°C with an R² of 0.956 across 60 km of cable, maintaining sub-degree accuracy in 88% of evaluation weeks (~96% where load was consistently recorded). Even across thermally distinct environments, onshore, ducted, and offshore, the model holds strong agreement between predicted and measured temperatures, with overall MAE under 0.56°C and near zero bias. This is an initial step towards using such methods to monitor cable thermal performance and conditions for infrastructure ~£1B. 

Recommendations for Critical Site investigation Information for Future NGET Projects 

The report provides a detailed strategy for the optimisation of desk-top studies of key thermal parameters which then informs the site investigation strategy for both offshore and offshore thermal characterisation of the cable route. Emphasis is also placed on the importance of mining extant DTS records to inform future builds. 

Final TRL – 6

The final TRL level of the project is 6: “technology model or prototype demonstration in a relevant environment”. This final TRL level is based on the following points:

1. User-friendly software has been produced that can display, and undertake analyses in real time, operational DTS data from any DTS system. 
2. Machine learning algorithms can accurately predict DTS temperature time series over annual timescales allowing for automatic detection of thermal drift or changes to environmental conditions. 
3. The FEA models, while requiring an expert user for initial implementation, are now able to be run (assuming appropriate software is installed) with a non-expert user editing an Excel sheet.  

Recommendations for further work
The AnTICS project primarily focused on submarine cable sections due to the availability of data. Significant developments are required to optimise onshore ground investigations to properly capture spatial and temporal variation of thermal properties. If sufficiently well developed, this could facilitate a reduction in the engineering of terrestrial trenches and broader installation costs.

As the project is finished at TRL6, any net benefits identified should be considered future opportunities, subject to further development, validation, and deployment beyond the scope of the current programme. In particular, the machine learning approaches developed within the project will require further validation, testing against expanded datasets, and demonstration of robustness across different environments and conditions before they can be relied upon for operational decision-making.
AnTICS has established the importance of correctly measuring the thermal properties of both marine and terrestrial soils. NG and UoS are now developing a proposal to assess the accuracy and precision of both laboratory and in situ approaches to measuring thermal conductivity/resistivity in both onshore and offshore scenarios, through a targeted field campaign at well-constrained test sites and hopefully including the EGL5 landfall site.

Lessons Learnt

Key Learning against Deliverable

D1 and D2 - The initial stages of the project focused on accumulation of relevant data, including DTS datasets, geotechnical information and cable installation information. A lesson learnt would be that standardised well documented storage of data would significantly accelerate delivery of projects. 

D4 and D5 - These deliverables demonstrated that a thermal modelling approach, with appropriate geotechnical information, can be used to assess anomalies in DTS data. This has implications for cable monitoring using DTS, such as the detection of cold spots or overheating. 

D7 - Demonstrated that applying machine learning to DTS data is powerful, but handling very large datasets requires significant processing and data management challenges to be overcome.

D9 – The significant impact of providing accurate, space and time-varying, thermal property data (ambient temperature, thermal resistivity and depth of cover), on cable rating and the importance of utilising well constrained DTS data for the validation of IEC / FE based models. Analysis of long-time series DTS data provides the maximum information on how cables perform so ensuring effective storage and utilisation of such data is paramount.

D11 – In this report the benefits of using extant DTS data to inform future installations were conveyed, along with the value to projects of publicly available datasets to inform geotechnical input parameters for cable thermal calculations. Guidance regarding conservative assessment of environmental thermal properties and measurements techniques was provided. 

D3, D6, D8, D10 and D12 – all update presentations / workshops. Positive discussions between NGET stakeholders and UoS staff assisted with the provision of data and provided insight into key opportunities to include findings in future cable systems installations. 

Dissemination of Findings

Workshops involving NGET staff took place as a structured part of the project in D6 and D12. 

A presentation of AnTICS outputs was made to the National Grid Offshore Engineering Team (an audience of both electrical engineers and geoscientists) held in Chester 09-10/09/25.

Academic dissemination:

• Estimation of Dielectric Temperature in Buried HVDC Cables Using Reduced-Order Thermal Model.         S. CHAUDHARY, J. Dix, P. Lewin, and G. Callender, Submitted in Proc. IEEE Int. Conf. Dielectrics (ICD), Southampton, UK, Jun. 21–25, 2026 (Abstract Accepted)
• Spatiotemporal Distributed Temperature Forecasting for High Voltage Transmission Cables with Temporal Fusion Transformers – CHAUDHARY, S., Dix, J., Lewin, P. & Callender, G. – Submitted (Under Review), IEEE Transactions in Power Delivery
• Visualisation and Analysis of Large Distributed Temperature Sensing (DTS) Datasets from High Voltage Cables - CHAUDHARY, S., Callender, G., Dix, J. & Lewin, P., 20 Oct 2025, Jicable HVDC'25 - 4th International Symposium on HVDC Cable Systems.
• Long-Term Seabed Temperature Measurements Derived from Distributed Temperature Sensing Systems attached to both Offshore Windfarm, and Marine Interconnector, High Voltage Cables: an In-Built, and Operating, Global Monitoring Network for Coastal and Shelf Seas. DIX, J., Chaudhary, S., Callender, G. February, 2026, AGU Oceans, Glasgow.