Break-point chlorination
Break-point chlorination — Break-point chlorination is the dose at which added chlorine stops being consumed by ammonia and organic matter and a free residual finally appears, so the plant's control target sits on the far side of a steep, strongly non-linear demand curve.
Contact time
Contact time — Disinfection credit is taken as the product of residual concentration and effective contact time at the worst hydraulic point in the basin, which is why a turbidity or residual sensor on a short-circuited path can certify a dose the water never received.
| Mounting | Where it is chosen | Dominant failure | What fixes it |
|---|---|---|---|
| In-line flow cell | Coagulant and polymer dosing lines | Coating build-up on the electrode because the line rarely goes dead | A clean-in-place tap and a scheduled brush cycle |
| Insertion probe in a pipe | Raw water and filter effluent | Stratification, so the probe reads a layer, not the flow | Move to the turbulent zone downstream of a bend or static mixer |
| Flow-through chamber | Filter effluent turbidity, final water | Air pockets trapped at the top of the chamber | Fill from the bottom and vent the highest point of the cell |
| Submerged probe | Basins, wet wells, reservoirs | Drying with residue on the sensor after a drain-down | Lift the probe before drain-down, or specify a wettable sensor |
| Turbidity window | biofilm and scaling |
|---|---|
| pH glass | junction clogging and dehydration |
| Redox electrode | surface poisoning, needs abrasion |
| Dissolved oxygen membrane | electrolyte level and diffuser fouling |
| Conductivity cell | coating in low-velocity branches |
| Level sensor in a wet well | foam and fats on the face |
- Illustrative drift of an unsupervised filter-effluent turbidity reading
- 0.4NTU over one fouling interval
- Source: Illustrative sample dataset, not a measured value.
Why does a perfectly calibrated sensor still control a loop badly in a plant?
Because the plant gives it a delay and a location that were chosen for hydraulics, not for control. A filter effluent analyzer may sit tens of seconds and many metres from the point where the coagulant dose can still change the outcome, so any controller reacting to it must be detuned, and the true fix is to move the measurement upstream or add a feed-forward term from the coagulant pump, not to tune the loop harder.
Is a submerged probe acceptable in a basin that is drained for cleaning?
Only if the sensor is designed to dry wet. A standard membrane or glass sensor that dries with plant water on it concentrates everything dissolved in that water onto the sensing surface, and the residue hardens into a film that survives the next immersion, so the sensor returns from the clean-out with an offset it never had before. The correct handling is to lift and rinse the probe before drain-down, or to specify a sensor whose membrane and reference system tolerate drying and re-wetting.
What does turbidity have to do with the chlorine contact basin?
Turbidity is the reason the chlorine demand curve moves: suspended particles shelter micro-organisms and consume residual, so credit for contact time is only valid while the turbidity at the inlet to the basin stays inside the range the credit was calculated for. That is why the turbidity measurement is a control input to disinfection and not merely a compliance reading, and why its mounting, its sample path and its cleaning interval all belong to the disinfection argument.
Every number on this page, including the intervals in the maintenance table and the drift figure in the stat block, is an illustrative sample value for a demonstration site. Replace them with your plant’s own records before publishing.
The loops that decide the specification
A treatment plant is a small number of control loops, and each one imposes a different requirement on the sensor.
Coagulant dosing is the loop where most instrumentation arguments are won or lost. The dose must follow raw-water quality, which changes with rain and with reservoir depth, and the only useful feedback is filtered water quality - which is late. So the working arrangement is a feed-forward term computed from raw-water turbidity and flow, trimmed slowly by filter effluent turbidity. The specification consequence is that the raw-water turbidity sensor needs a cleaning interval that survives the dirtiest week of the year, because its reading drives the dose continuously, and an alarm on cleaning failure is more valuable than another decimal of resolution.
Aeration is the largest energy consumer, and the dissolved oxygen loop sets the trade between energy and effluent quality. The sensor here is not a measurement but a control input to blowers or valves, which makes drift expensive in a different currency: an oxygen sensor drifting low makes the plant over-aerate and pay for air it does not need, and the drift usually comes from the diffuser or the electrolyte, not the electronics. Mounting in a mixing zone, away from the bubble plume, and a fouling-tolerant diaphragm matter more than accuracy in the certificate.
Chlorination is the loop where a sensor is a legal statement. Residual and turbidity at the end of the contact path support the calculated credit, so the measurement has to be representative of the worst-diluted portion of the flow. Where the hydraulics short-circuit, no instrument can compensate, and the honest answer is a mixing improvement, then a sensor.
Sample points that mislead
A surprising number of plant readings describe pipework rather than water. Three patterns recur.
A dead leg: a branch off a main that carries almost no flow, into which a sensor has been inserted because that is where the tap existed. Water in that branch is old and stratified, its temperature and residual differ from the main, and the reading is stable and wrong, which is worse than a noisy reading. The test is simple - open the tap and see how long flow takes to establish, and whether the reading changes once it does.
An air pocket: an upward-facing sensor body in a line that fills intermittently, or a chamber filled from the top. A trapped bubble sits on a turbidity window like a lens and on a pH bulb like a blanket, and it produces offsets that come and go with the day’s duty cycle. Fill chambers from the bottom, keep the cell the highest point of its own vent path, and slope the sample line continuously.
A velocity problem: conductivity and pH cells need flow to keep their surfaces clean, but they are frequently installed in the low-velocity branch of a tee where the coating grows fastest. If a low-flow condition is unavoidable, the maintenance plan has to assume the coating, and the plan beats the calibration.
Failure modes seen repeatedly in the field
Biofilm is the default, not the exception: on any window or membrane in treated or raw water, a biological film establishes within days and its growth rate depends on temperature, nutrients and surface material. It reads as a slow monotonic drift upward in turbidity, with a characteristic failure to return to zero after a chemical clean because a film residue remains. Cleaning frequency, mechanical wipers or a UV-LED source with a reference channel are responses to that biology, and all three must be specified against the warmest month.
Dry-out is the second. Plants drain basins and filter boxes, and the sensor stays in. Residue on glass and membranes hardens; some reference electrolytes crystallise; some membranes change their oxygen transfer after drying. The remedy is procedural as much as technical, so the commissioning documentation should state whether the instrument may stay in during drain-down.
Chemical attack follows the cleaning regime rather than the process: shock chlorination and cleaning with acids or alkaline detergents is far more aggressive than the water being measured, and it hits polycarbonate windows, cable sheaths and gaskets. If the plant cleans in place, choose the materials against the clean, not against the sample.
Electrical noise arrives from the equipment around it - pumps with variable-speed drives in the same wet well, sharing a grounding arrangement with the instrument loop, whose high-frequency leakage appears as a moving offset on a high-impedance pH signal. This is why analogue high-impedance signals are run on shielded cable with the screen referenced at one end only, and why an unexplained offset that correlates with drive speed is an earthing problem rather than a sensor problem.
Selecting alongside the maintenance plan
Every instrument in a plant is eventually maintained by a person with a bucket, a brush and a limited window, so the specification that survives is the one that fits that window: a sensor that can be withdrawn without draining the line where that is possible, a cleanable optical window reachable without a tool, spares for membranes and electrolyte, and a trend channel that makes drift visible to the operator before it reaches a control loop. If the plant has no calibration records, assume the drift and design the process tolerance around it, because the alternative is that a slow fouling curve quietly becomes the operating limit of a filter.