Increasing automation and confidence in non-revenue water workflows with SensorClean

The global direct cost of non-revenue water (NRW) is 39 billion US dollars per year (Liemberger and Wyatt 2019).

In large cities, reducing a small percentage of water loss can save tens of millions of dollars by deferring capital investment for new sources of water, such as desalination plants.  

NRW is a significant financial loss, uses valuable maintenance resources and doesn’t bode well in times of extreme drought and climate change.  

To inform maintenance strategies, spreadsheets are the most popular tool for analyzing NRW and customer big data sets.

But big data tools on the cloud are built for big data analysis.

Increasing automation and confidence with SensorClean

Engineers who use SensorClean can easily find and quantify different types of NRW for their District Metered Areas (DMAs).

For example, the figure below shows hydrant water loss that was easily found and quantified.

Figure 1 Identified hydrant flows in SensorClean

Revenue hydrant extraction with known dates can be deducted from DMA analysis.

Organizing NRW events in SensorClean also saves time with other DMA use-cases. For example data-driven capacity assessments that can potentially save millions of dollars in Net Present Value analyses (see blog).

Significant big data problems with spreadsheets

For a water-year, a flow sensor may record half a million observations (say 1 per minute). Let’s say the utility has 200 sensors for its DMAs.

Engineers will then be analyzing 100 million rows of data in spreadsheets to understand NRW spatially.

But an Excel Worksheet allows only 1 million rows, so an inefficient big data workflow occurs.  

Current practice is usually to adopt flow rates from guidelines instead of using the actual system data.

Using real data gives confidence in the analysis.

Where is NRW occurring (bottom-up approach)?

The goal is to locate and quantify different types of NRW and then rank issues for maintenance priority.

In general, the different types of NRW – which require different maintenance strategies - are:

1. Large pipe failures – usually known to utilities

2. Low-level leakage – harder to find

3. Hydrant flows – some revenue, some theft

4. Valve water loss, usually from maintenance; and

5. Other losses from data errors and tanks with leaks or that overflow.

What is the total NRW (top-down approach)?

This estimates the difference between what was going in and out of the overall system for a financial year or ‘water-year’.

Inflow and outflow are estimated with source flow and customer meter data respectively.

As customer meter data is recorded by staff walking around neighborhoods - rather than an instantaneous digital reading - a misalignment of the inflow and outflow times occurs, introducing error.  

Many utilities are slowly installing digital meters to reduce this error and to reduce leakage by informing customers about unusually high flows.

Conclusion

Water engineers using SensorClean save time and money and have increased confidence in NRW analysis to prioritize maintenance.

Please contact us to learn more about SensorClean.

References

Liemberger, R. and A. Wyatt (2019). "Quantifying the global non-revenue water problem." Water supply 19(3): 831-837.