Learnings

Outcomes

During the reporting period from April 2025 to March 2026, the HV Pilot project has delivered a range of outcomes that demonstrate strong progress towards addressing the identified gap in high-voltage network connectivity data.

The project has successfully moved beyond initial investigation and has now developed and demonstrated a working approach using real network data. While the project is still ongoing, the outcomes to date show that the method is both technically viable and practically applicable.

Demonstrated ability to identify network connections

The most significant outcome is that the project has shown it is possible to identify the phases of single-phase transformers using existing data sources.

• Smart meter data has been used to infer how customers and equipment are connected across multiple feeders
• Strong and consistent results have been achieved across a range of network scenarios, particularly on simpler feeders
• Results have been validated using multiple independent analytical approaches, increasing confidence in outputs
   
  This represents a clear improvement over current approaches, which rely on incomplete records or manual surveys.

Development of a practical methodology

The project has produced a working and repeatable methodology for identifying connectivity.

• Multiple analytical approaches have been developed and tested
• A combined approach has been established to improve reliability
• A structured process has been defined to select the most reliable result

This provides a practical framework that can be applied beyond the trial environment and adapted for different network conditions.

Improved understanding of network data

The project has significantly improved understanding of existing network data and its limitations.

Data inconsistencies and gaps have been identified across multiple systems
The impact of data quality on analytical outcomes has been clearly demonstrated
A structured approach to managing and combining data has been developed
 

This outcome supports improved network modelling and planning beyond the scope of this project.

Development of data infrastructure and analytical environment

The project has also delivered a structured data and analytics capability to support the solution.

• A dedicated HV Pilot datastore has been developed to manage network and smart meter dataset.
• An Azure-based analytics environment has been established to enable scalable data processing and analysis
• Data models and integration processes have been developed to align and combine multiple data sources
   

This infrastructure has been designed with scalability in mind, supporting potential transition towards business-as-usual deployment. While further development and integration would be required for full operational rollout, it provides a strong foundation for future use.

Assessment of supporting technologies

The project has assessed the role of aerial imagery, LiDAR and machine vision.

• Imaging techniques can identify certain network features and support validation
• However, the number of usable images and data quality limitations reduce effectiveness
• Imaging has been established as a supporting tool rather than the primary method 

This provides a clear understanding of how these technologies can be applied in future solutions.

Identification of performance drivers

The project has identified the main factors that influence performance.

• Network complexity affects consistency of results, with simpler networks performing more strongly
• Data quality and completeness are key drivers of accuracy
• Specific network equipment, such as voltage regulators, can influence outputs
   

Understanding these factors enables targeted improvement of the method.

Progression in Technology Readiness

The project has made clear progress in developing the method beyond its initial stage.

• At the start of the project, the approach was at an exploratory stage and largely unproven in practice
• The project has now demonstrated the method using real data across multiple network scenarios
• The solution has progressed to a stage where it can be considered technically demonstrated in a relevant environment

The project has progressed from an exploratory stage to demonstrating the method using real-world data in a trial environment, representing a clear advancement in its Technology Readiness.Further work is required to confirm performance at scale and prepare for operational deployment.

Performance improvements

The outcomes demonstrate measurable improvements compared to existing approaches.

• Significant reduction in the need for manual surveys to identify network connections
• Ability to analyse large volumes of network data efficiently using existing systems
• Increased confidence in network connectivity information where data quality is sufficient

These improvements highlight the potential for operational efficiency and cost reduction.

Opportunities for future development

The project has identified clear opportunities for future work to develop learning further.

• Extending testing across a wider range of networks and operating conditions
• Improving performance in more complex network areas
• Enhancing data quality, availability and standardisation
• Improving confidence scoring and decision-support outputs
• Integrating the method into existing operational systems

These opportunities provide a clear basis for future innovation projects and further development.

Pathway to larger-scale deployment

Based on the outcomes to date, there is a clear pathway for further development and potential deployment.

• The smart meter-based method has strong potential for wider deployment where suitable data is available
• A hybrid approach combining multiple data sources offers the most reliable solution
• The method is likely to be used initially for targeted applications before wider rollout
   

A follow-on trial is recommended to demonstrate performance at larger scale, support integration into operational workflows, and confirm the business case.

Effectiveness of the project

Overall, the project has been effective in delivering its intended research and development outcomes.

• A practical solution to a known data gap has been demonstrated
• Multiple approaches have been tested and compared
• Real-world challenges have been identified and addressed 

The project has successfully progressed from concept to a working method, providing a strong foundation for the final phase and future development.

Summary of outcomes

In summary, the project has:

• Demonstrated that high-voltage network connectivity can be identified using existing data
• Achieved high levels of accuracy in many scenarios
• Developed a practical and repeatable methodology
• Delivered supporting data infrastructure and analytical capabilities
• Improved understanding of network data and its limitations
• Identified the most effective approach and supporting technologies
• Established a clear pathway for further development and potential deployment 

These outcomes align with the original aims of the project and provide a solid basis for progressing to the next stage.

 

Lessons Learnt

The HV Pilot project has generated a range of practical insights during the reporting period April 2025 to March 2026. 

These lessons reflect the progress made in developing and testing methods to identify high-voltage network connectivity and will inform both the final stage of this project and future work in this area.

Data availability, quality and access

A key learning from the project is that data availability alone is not sufficient. While relevant data exists across multiple systems, the project has shown that data is often inconsistent, incomplete or misaligned.

Future projects should prioritise:

• Establishing clear data governance and ownership
• Standardising data formats across systems
• Implementing automated processes for data cleaning and alignmentImproving data quality will be essential for scaling the method successfully.

Smart meter data provides the core solution

The project has demonstrated that smart meter data provides the most reliable method for identifying network connections.

• High levels of accuracy have been achieved in many cases
• Results improve when multiple analytical methods are compared
• The approach reduces reliance on manual surveys
• Future development should focus on refining and scaling this approach as the primary solution.

Performance varies depending on network characteristics

Performance is influenced by network type and complexity.

• Simpler networks give more consistent results
• More complex networks require additional consideration
• Equipment such as voltage regulators can affect outcomes
• Future methods should incorporate adaptive approaches to handle different network conditions and include clear confidence measures for users.

Imaging and machine vision support the main method

The project has shown that imaging can support analysis but is not currently suitable as a standalone solution.

• It can assist with validation and investigation
• Data quality limits its reliability
• It requires improved data capture to be more effective
• Future work should focus on targeted use of imaging to support smart meter analysis rather than replacing it.

Combining methods improves confidence

A key outcome is that combining multiple analytical approaches improves reliability.

• Different methods provide complementary insights
• Agreement between methods increases confidence
• A hybrid approach produces the most robust results
• Future implementations should adopt this combined approach as standard practice.

Data engineering is a critical enabler

The project has shown that significant effort is required to prepare and manage data, and that data engineering is a key part of the overall solution.

The HV Pilot project has developed a structured data architecture to support this work, including:

• A central datastore for network and smart meter dat
• Data ingestion and processing pipelines

Data integration processes to align multiple datasets

These capabilities have enabled reliable analysis and demonstrate that the project has delivered not only algorithm development, but also the data foundations required for future operational deployment.

Future projects should include dedicated focus on:

• Data ingestion and processing pipelines
• Ongoing data validation and updates
• Clear processes for managing anomalies

This will be essential for scaling the method and supporting business-as-usual implementation.

Practical deployment considerations

The project has provided insight into how the method could be used in practice.

• The approach is likely to be used as an on-demand tool
• It can support targeted operational decisions
• It reduces reliance on manual field surveys

To enable this, future deployment will require integration with existing systems, defined processes and user guidance.

Recommended next steps and trial development

To progress the method to the next stage of maturity, a follow-on trial is recommended.

This should focus on:

• Deployment across a larger and more diverse set of networks
• Testing performance under a wider range of operating conditions
• Integration with operational systems and workflows
• Validation of results at scale
• This will support progression to a higher level of Technology Readiness and confirm suitability for wider adoption.

Likelihood of large-scale deployment

Based on the outcomes to date, there is a strong likelihood that the method can be deployed more widely, particularly where suitable smart meter data is available.

However, successful large-scale deployment will depend on:

• Improvements in data quality and consistency
• Demonstration of performance across a broader range of networks
• Integration into existing operational systems
• With these conditions met, the method has clear potential for wider application.

Effectiveness of the research and demonstration

The research and development undertaken in this project has been effective in progressing from concept to a working solution.

• Multiple methods have been developed and tested
• A practical and repeatable approach has been identified
• Real-world challenges have been clearly understood
• The project has delivered meaningful learning and established a strong foundation for further development.

Overall learning

Overall, the project has shown that it is possible to identify high-voltage network connections using existing data.

It has confirmed that:

• Smart meter data provides the most effective solution
• Combining methods improves reliability
• Data quality is critical to success
• These learning points provide a clear basis for future work and support progression towards operational deployment.