What good CSO monitoring actually looks like

A priority framework for utilities standing up combined sewer overflow monitoring programs.

Utilities running combined sewer systems (CSOs) are under more pressure than they've been in decades. Consent decrees, EPA reporting obligations, and Long-Term Control Plan milestones have moved CSO monitoring from a nice-to-have to a regulatory necessity.

The question for most utilities is how to do it well enough that the data holds up to a regulator, an auditor, and the operations team that has to use it every day.

We've been asked a lot lately what good CSO monitoring actually looks like.

Here's how we'd answer that, in order of what actually matters.

A CSO being monitored with Kallipr Captis radar devices.

The things that have to be right, or the program doesn't work

Accurate overflow detection.

The system has to prove a spill actually occurred, not just that the level was high. That often means more than one sensing point, or smarter logic to separate true spills from surcharge and backwater conditions. A high reading is not an overflow, and a regulator will eventually ask you to demonstrate the difference. Programs that conflate the two end up either over-reporting (which creates compliance exposure the utility didn't actually earn) or under-reporting (which is worse).

High-frequency logging and transmission.

Compliance reporting needs precise start and stop times for every overflow event. That means down to 1-minute or sub-minute logging during active conditions, and fast alarm transmission while the event is still happening. Daily uploads after the fact don't meet most modern reporting requirements, and they certainly don't give operations any chance to respond in real time.

Sensors that survive surcharge and turbulence.

Overflow structures see full-pipe conditions, debris, and rapid level changes during the exact events the program is designed to capture. The sensor has to stay accurate through all of it. A device that drops out during surcharge is a device that misses every event worth recording.

Multiple sensors where the hydraulics demand it.

Some overflow points need both wastewater and stormwater levels monitored to confirm a discharge actually happened. A single sensor in a tidally-influenced or flood-prone location will generate false overflow alarms during downstream flooding, which you'll spend the next twelve months explaining. The hydraulic complexity of the site should drive the monitoring design, not the other way around.

Kallipr devices can connect to multiple radars or sensors, making them ideal for CSO monitoring

The things that determine whether the program is trusted long-term

Communications and power resilience.

Overflow events happen during storms, which is also when cellular coverage and grid power are at their worst. Devices need genuine battery capacity, local data buffering so nothing is lost during transmission gaps, and communications that hold up through extended wet weather. A monitoring program that goes dark during the event it’s supposed to be monitoring is pretty useless.

Maintainability and battery strategy.

CSO monitoring runs on a different power curve than standard sewer level work. Sub-minute logging during active events, frequent alarm transmission through wet weather, and high-volume data buffering all burn battery. Realistic battery life on a high-activity CSO site is closer to one to three years than the 5+ years a low-activity application might see. What matters in a CSO program is not how long the battery lasts in theory, it's how quickly and safely the field crew can replace it when the time comes. Field-swappable batteries, tool-free access, and an install design that doesn't require removing the device from the chamber or exposing electronics are what make a CSO monitoring program operationally sustainable across a hundred sites.

Alarm logic that doesn't cry wolf.

Turbulence and short surcharges will generate nuisance spill alarms unless the system uses dwell times, hysteresis, and multi-reading validation. Nuisance alarms train operators to ignore the platform, and once that pattern sets in, the real alarms get missed too. Alarm logic is a determinant of whether the program is actually used.

Compliance-grade data and audit trail.

Telemetry data ends up in regulatory submissions, consent decree reporting, and occasionally in litigation. That means secure storage, accurate timestamps, and event history that holds up under audit. Good enough for operations is not good enough for a consent decree, and the gap between the two is where utilities get caught out.

Custom brackets that can be altered to suit the CSO conditions are important for install ease

The things that won't sink the program but will quietly cost you

Installation access and safety.

Overflow structures tend to be deeper, more hazardous, and harder to access than standard manholes. Traffic management, confined space entry permits, and flood risk all add real cost to deployment, and that cost compounds across a hundred-site rollout. Install design choices made on day one ripple through the next decade of maintenance.

Maintenance and cleaning requirements.

Debris and fouling drive more frequent inspections at overflow sites than anywhere else in the network. A poor install design like a sensor in the wrong position, mount that traps debris, or a cable that catches rag will multiply that workload for the life of the asset.

Chamber and structure condition.

Older overflow structures often have corrosion, damaged weirs, or undocumented layouts that affect sensor placement and long-term reliability. The structure itself is part of the monitoring system. Survey it properly before you spec the install, not after.

Why the order matters

The reason we present this as a priority framework rather than a flat checklist is that the failure modes are different at each tier. Get the top tier wrong and the program produces compliance exposure. Get the middle tier wrong and the program stops being trusted, and stops being used. Get the bottom tier wrong and the program runs over budget for the rest of its life.

Most CSO monitoring failures we've seen weren't caused by the sensor. They were caused by a program design that didn't think hierarchically about what mattered most.

If you're standing up a CSO monitoring program, the question isn't which device to buy. It's which of these twelve things you've got a credible answer for. The device choice follows from that, not the other way around.