Reverse Osmosis systems are one of the more challenging use cases due to variability in event frequency, flow rate, duration and gallons consumed per water production cycle. Water conditions (TDS, temperature, pressure) and system design (membrane capacity, membrane condition, storage volume) do not permit use of a universal leak identifier. After the Smart Water Shutoff has been installed it goes through an initial learning period to characterize normal system use. System data is periodically analyzed to update the model for the system.

In most installations, the Reverse Osmosis system events are identified during this learning process. In the event a false-positive is triggered by the Reverse Osmosis system there is a re-training regime that will incorporate the event information to update the leak detection algorithms.