Benchmarking residential water use

I have been experimenting in my electoral area with benchmarks for utility use (in this case: water).  We have water meters; however, our utilities software is not particularly good at generating information for water customers.  So we created dashboards that draw data from our billing software and present the information in a more accessible way.

In our experience, ratepayers have three main questions about water:

1. Why is my bill so high?
2. What does my water use look like over time?
3. How does my water use compare to that of my neighbors?

High bills

High water bills are a function of the cost of the utiltiy and the allocation of those costs to users.  Managing the cost of the utility is critical, which is one of the reasons we use Benchcube dashboards to identify cost drivers and use benchmarks to identify potential best practices in other communities.  Allocation of these costs in a metered environment comes down to water use.

Water use over time

Water use over time is interesting--especially in arid regions--due to the massive difference between summer and winter use.  The bar graph below shows monthly water consumption of time for one of my constituents. We can focus only on the blue bars at this point:

Two things jump out at me in this graph:

1. It does not take much water to wash dishes, toilets, take showers, do laundry, and so on.  All these activities occur in November-March.  But we can see that the water requirements in winter are a small relative to the summer months.  Summer mean lawns, gardens, shrubs swimming pools, and other uses.  These are large (and mostly discretionary) uses of water.
2. This resident saw a large summer-to-summer increase in water use, which translated into an increase of almost $100/month.  In this case, the culprit was an accidental change to an automatic irrigation controller. 

Benchmarking water use

The lines on the graph provide some context for the resident's water use.  Specifically, we took the monthly water use for all users within the same rate class (residential, agricultural) and determined three values:

1. Light use: This is the level of water use that defines the bottom decile of utility customers.  In other words, 90% of properties use more water than this.
2. Typical use:  This is the group median water consumption.
3. Heavy use: This is the level of water use at which 90% of properties use less.

So in this case, the resident in the graph above was close to the median in 2015 but shifted to the top decile in the summer of 2016.

Another way to look at this distributional information is as a histogram.  The graph below shows the distribution of monthly consumption (per m2) for residents.  The dot shows the resident's consumption during the month.  This view simply complements the high-medium-low bands on the bar chart above.  One interesting feature of this distribution is that water use appears not to be normally distributed.  The blue dot is clearly on the fat-tail end of the distribution, but not at the far right. The highest bar is in the neighborhood of 4 l/m2.  However, some residents are using 10x this amount of water per square meter.