The VM Data project will deliver a VM capability across our Low Voltage network. This will reduce need for physical monitoring and improve our knowledge of asset loading against time, thus avoiding the costs associated with physical monitoring and demonstrating RIIO-ED2 cost savings and transition to Distribution System Operator.
Current lack of access to half-hourly data about household power flows on our network inhibits the understanding of LV network load flows, and of where electric vehicles and low carbon technologies are connected at LV level. With the acceleration of LCT take up, this could result in clustering on the network which then creates a need to install physical monitoring at substations to monitor the loading of the network. The VM Data project will investigate the feasibility of creating half-hourly load profiles for WPD’s customers, including those with EV / LCT that can be fed into a Virtual Monitoring tool for the LV networks.
Benefits
Providing virtual LV network monitoring and improving the accuracy of our records for LCT installations will result in greater confidence around the actual capacity of the network. This will have a knock-on impact on customers seeking increased capacity or new connections. A more accurate assessment of existing capacity will allow for a more accurate assessment of potential reinforcement activities ensuring fairness between the customers benefitting from the capacity and the general customer base. An understanding of the connected PV capacity is useful to help control engineers determine the degree to which the network loads are offset by embedded generation.
Furthermore, better understanding of LV network loading and capacity will lead to improved network reliability for existing customers and faster customer connection times for new customers.
Learnings
Outcomes
In addition to the learning captured in section 6, this project established:
- There is value to being able to predict substation profiles form customer HH data.
- The analytics environment that was required to deliver the required outputs.
- A number of datasets as identified in the discovery phase, including;
- Anonymised or pseudo-anonymised HH MPAN data
- Energy Performance Certificate
- LCT presence
- Substation demographics
The full suite of reports developed by this project can be found at www.westernpower.co.uk/Innovation/projects/virtual-moniotring-data-vm-data.
Lessons Learnt
This project undertook a number of analysis sprints that conducted works to inform final process design. This led to the following technical learning points about the WPD network and application of analytics:
- Algorithm for profile identification. During the project some analysis work was undertaken to assess different clustering algorithms and establish which are the most suitable for the purpose of identifying discrete profiles from half hourly consumption. This compared a number of different algorithms and it was concluded that Hierarchical methodology gives the best results.
- Seasonal and Weekday/Weekend Profiles. Analysis was undertaken to determine whether the distinction between weekday and weekend is an important factor in defining the profile shape. The conclusion was made that there needs to be a distinct set of profiles to allow for weekends and that a “one profile for all time” does not give acceptable outcomes.
- Propensity to change supplier. As part of this data analysis, an MPAN’s propensity to change supplier was defined. This metric is an additional attribute which can be used in future analysis in both the LCT detection and consumption profile workstreams. This metric was calculated for 70% of WPD MPANs. It was also discovered that there is no evidence that customers with LCT are more likely to switch suppliers.
- Correlation between LCT ownership and Energy Supplier. To identify attributes that would be useful in the LCT detection model analysis was undertaken to test whether there would be a correlation between LCT ownership and certain energy suppliers. It was concluded that customers of the ‘Big 6’ suppliers are generally less likely to have LCT. The exception to this is Scottish Power for both EV and PV and SSE for PV only.
- Relationship between Energy Performance Certificates and ADC
Analysis was undertaken to identify attributes that indicate different patterns of energy consumption. A number of factors were found to be related to different ADC summer to winter ratios, which included:
- Main heating energy efficiency
- Hot water energy efficiency
- Presence of mains gas
- Relationship between building fabric and energy consumption or LCT installation.
- Properties with LCT are more likely to be houses, particularly detached houses for EV.
- Bungalows are more likely to have PV than the population as a whole.
- The number of rooms does not influence the pattern of energy usage, but it does influence the absolute level of energy consumption.
- Substation properties. The number of rooms is highly correlated to the amount of energy consumed, but not to the difference between summer and winter consumption.
- Substation estate. To determine the extent to which a better understanding of domestic consumption patterns can aid in understanding the load at the distribution substation level, analysis was undertaken to set out likely sensitivity parameters for domestic demand on distribution substations. This exercise did develop new learning insofar that it was found that:
- 25% of substations supply only domestic MPANs and for 50% of the substations >90% of the connected MPANs are domestic MPANs. It was also observed that 20% of substations supply only commercial MPANs
- 95% of distribution substations have no known EV connected
- 72% of distribution substations have no known PV connected
- Almost 50% of distribution substations have no smart meters
