In the water industry, monitoring and reporting are often based on estimations and extrapolations. This involves measuring quantities from samples, indirect estimations, or modelling. Even with direct measurements, such as measured consumption data, the resolution or coverage might be insufficient, necessitating further modelling for certain applications. When it comes to consumption and demand variables, key measurement methods include property monitors, area monitors, and smart metering, each with its own set of limitations.
In AMP8 we expect to see a step change in the number of smart meters being installed. These have the potential to improve the accuracy monitoring and reporting. With the smart meter rollout, we also expect to see more unmeasured properties being metered and moving onto a metered bill. This will change the characteristics of the unmeasured population, meaning that existing estimations and modelling methods will need to adapt to these changes. Examples of this include the measurement of unmeasured household consumption, for which key measurement methods include property monitors, area monitors, and smart metering.
Navigating Multiple Information Sources
How do we manage these diverse information sources and steer towards a future increasingly dominated by smart metering? At Artesia, we have developed and implemented a solution based on Data Assimilation. This allows the increasing availability of smart meter data to feed into monitoring and reporting systems alongside the existing methods and approaches. The Data Assimilation concept, borrowed from meteorology, integrates various sources of information by weighting them according to their relative uncertainty. The more a source, the greater its weight. This method is applied on a stratified basis, ensuring that each source weights more where it is most representative and reliable.
Advantages of Data Assimilation
The approach offers numerous benefits:
- Resilience: By incorporating multiple information sources, the system remains functional even if one source degrades, becomes unavailable, or highly uncertain.
- Future-Proofing: By adjusting weights over time, the system can adapt to improvements in some sources (e.g., smart metering rollout) and the decline of others (e.g., monitor loss of representativeness).
- Gradual Adaptation: This method allows for continuous adjustments without abrupt variations and step-changes.
- Sense Check: Significant deviations of one of the estimations compared to the others, outside uncertainty bands can be easily detected, indicating potential issues.
- Reduced Need for Methodological Changes: working with weights, most changes can be assimilated as data changes rather than methodological changes, reducing the need for formal methodological updates across the Asset Management Plan cycle.
- Better Understanding of Uncertainties: The methodology enhances understanding of uncertainty factors and the weaknesses of each estimation.
- Stratification: By applying the methodology on a stratified basis, we can use each source of information for what it excels at. For example, a monitor representing urban properties well but poorly representing rural areas will have stronger weights for urban properties and weaker weights for rural ones.
Looking Ahead
What’s next? We are working on improvements to better handle non-normal distributions and data scarcity. Normal distribution assumptions are common in the water industry due to simpler mathematics, but often they don’t reflect reality. We are collaborating with academic partners to develop solutions that account for the skewed, asymmetrical distributions often seen in consumption and flows. This challenging work aims to provide more realistic estimations and address biases in current methodologies.
Get in Touch
If you are interested in exploring how data assimilation can address your challenges, please get in touch with us at Artesia. We are committed to supporting the industry with innovative, accurate, and resilient solutions.
