Waste Water Pump Station Monitoring


  • Situation
    • Sewerage systems get blocked, clogged, overflow and send untreated waste into an estuary used for commercial oyster cultivation.
    • Limited data from sewerage system has historically made blockage detection challenging and dependent on local residents calling water utility.
  • Goal is to detect events without installing extensive new sensor networks throughout the entire sewerage system network


  • Develop a standard weekly usage profile for each pump out station using reference profile
  • Utilize asset modeling to take the single model and create a unique usage shape for each individual pump station
  • Monitor all pumping stations for large deviations from “normal” profiles
  • Integrated with notifications to send alerts to operations center


  • Fast deployment (in less than 1 month) giving improved early detection of blockages
  • Additional capital investment deferred
  • Seeq model and notifications identify abnormal conditions in real-time (Multiple network operating anomalies successfully detected)
  • Engineers can tune the model easily
  • Rich data context for abnormal events

Data Sources

  • OSI PI
  • PI Asset Framework

Data Cleansing

  • Transform pump on-off signals to a more meaningful analytic – time between cycles – to create a virtual ”sewage load” signal

Calculations and Capsules

  • Developed capsules to estimate normal pump run frequency
  • Identified rain events and eliminated those times from the model
  • Used to detect unusual pumping patterns due to blockages

Summarizing Results

  • Rapid development in less than 1 month
  • Blockage detected 13 hours prior to reporting
  • Delayed capital investment
  • Reduced frequency of release of untreated sewage:
    • Damage local oyster producers business
    • Pollution and potential clean-up costs avoided
    • Water company reputational harm avoided