To model the uncertainties underlying changing demands on the network and therefore the need for load-related investment.
Benefits
- Establishment of external influences on the need for network development – generation and demand including installation of solar PV – can be modelled probabilistically in an efficient manner providing useful information to network development planners and network designers on network performance under different conditions.
- Evidence to support a decision on whether or not to engage with a professional software provider to develop a production grade, supported software tool enabling probabilistic network planning.
- Specification for suitable professional simulation and analysis tools to exploit models of external influences on network adequacy and for a tool to interrogate and analyse its outputs.
Learnings
Outcomes
The following outcomes of the project were realised:
- The project resulted in the development of a prototype tool for the probabilistic analysis of network performance under uncertain conditions. Although this prototype is not (and was not intended to be) of “production quality”, its potential to support the network assessment tasks required as part of price control preparation has been evaluated and tested.
- The use of the prototype software, methods and techniques developed as part of the project to support network analysis activities associated with long-term reinforcement planning in price-control timescales has also been evaluated.
- SPEN’s university project partner completed their assignment but SPEN has determined that we cannot use the product. SPEN is planning maximum contracted rating for the status quo. However, the project concept is that this approach is over-engineering which stops new connections.
Lessons Learnt
The following lessons were learnt as part of the project:
- Methods for the statistical representation of LCT behaviour – by decomposing the complex overall behaviour into a series of relatively simple models which describe behaviour at different timescales – appeared to show promise for the generation of new representative time series to represent future behaviour, or behaviour at different locations in a network.
- Graph-based modelling approaches were effective in facilitating manipulation and analysis of distribution network connectivity and its relationship to generation and load.
- A mixture of statistical and direct historical modelling and sampling methods were applied; it was found that the appropriate choice depended on the nature of the quantity being sampled and the availability of appropriate historical data (e.g., for weather data geographical relevance is important).
- The selection of a broad range of test cases was useful in ensuring wider applicability of the toolset: examples were considered from the transmission boundary to 11kV level. The set of required analysis methods and network adequacy and security criteria varied across this range and has required additional capabilities to be developed in comparison to the use of test cases using similar networks.
- Comparison of the performance of the developed toolset with existing “Business as Usual” methods across a range of network operator activities (including small-scale trials) is recommended to assess appropriate ways in which the methods, software tools and learning from the project can be best exploited. The capabilities of the prototype relative to existing software tools in use within the business should be assessed and compared.
- In the light of this comparison, a decision should be taken as to whether development of the prototype (and its embodied methods) into a “production quality” tool, should be pursued in collaboration with suitable partner organisations. Alternatively, it may be possible to incorporate elements of the prototype (and its methods) into existing workflows as “extensions” to existing tools.
- A clear lesson to date is the importance of a comprehensive survey of relevant existing learning at an international scale, including both utility-led innovation and academic literature.