COD Sensor Sampling: Reject Stale Water After Pump Stops

Quick Answer: COD sensor sampling is unreliable when a bypass stops delivering current process water, even if the probe continues returning plausible numbers. Use a separate indication of sample delivery, exclude stagnant-water periods from normal process interpretation, and verify flushing after restart. A pump command or a successful Modbus response does not prove that fresh water has reached the optical path.

The sensor can be working while the sample is stale

A stopped pump, blocked intake or closed valve can leave water around an immersed optical probe. The instrument may still communicate normally because it remains powered and wet. Its reading then describes the trapped sample, not necessarily the current plant discharge or tank condition.

The installation question is therefore more specific than whether to choose a tank or cabinet: what confirms delivery to the measurement point, and how will the records be labeled when delivery is lost? The same issue applies to a conductivity probe in the shared sample path; a stable EC value is not proof of flow.

Track COD sensor sampling through four operating states

These are proposed controller states, not claimed built-in functions of the NBL-WQ-COD. A PLC, RTU or data logger must implement them with suitable inputs. Select a sample-delivery indication appropriate to the assembly and test the failure modes it can and cannot detect. A running pump can still face an empty intake or a downstream blockage.

Check other sensors on the same path individually. A pH probe might remain immersed in retained liquid while a different part of the assembly drains. Shared pump status does not establish immersion, representative sampling or readiness for every channel.

Estimate residence time, then validate restart behavior

For a simple screening calculation, divide the retained water volume by the measured delivery flow: nominal residence time = volume / flow. For example, 3 L of retained volume at 0.5 L/min gives 6 minutes. This is an illustrative calculation, not an approved flushing time or an actual project result.

Include the intake tube, fittings, vessel and other significant retained volumes. Mixing, dead legs, variable flow and deposits mean that exchanging one nominal volume does not prove complete replacement. Measure delivery at the relevant operating condition and establish the required recovery period through a controlled commissioning test.

Record the time at which delivery resumes, when newly delivered water reaches the probe and when readings become suitable for the intended decision. Take comparison samples at the same measurement location and align them with that time basis. For the broader layout choice, see direct immersion versus a sample bypass.

The PH-4S probe shown below requires its own installation limits and service checks. When several instruments occupy one vessel, the water path and mounting arrangement must satisfy all of them, rather than treating the vessel as an unrestricted universal holder.

Apply the actual COD probe limits to the sample assembly

The NiuBoL NBL-WQ-COD is an immersion instrument using dual-wavelength UV absorption. Its turbidity compensation addresses part of the optical interference; it cannot correct a stale sample, and it does not establish equivalence to every laboratory COD method.

If the station also measures dissolved oxygen, consider whether the sample route changes the oxygen condition before it reaches that probe. The separate DO-4A instrument below cannot establish that the COD sample represents current discharge, nor can a valid COD record validate the oxygen channel.

Project Decision FAQ

Q1. Why is COD still changing after the sample pump stops?

The sensor may be reading retained water whose condition is changing, or its optical surface may be affected by bubbles or deposits. Check sample delivery and probe condition before interpreting the change as current process behavior.

Q2. Does pump-running feedback prove fresh sample delivery?

No. It confirms an equipment state, not necessarily water arrival. Verify delivery through an appropriate independent check and include blocked-intake, closed-valve and low-water conditions in commissioning.

Q3. Is one calculated residence time a sufficient flush?

Not necessarily. The volume/flow calculation is an estimate. Mixing and stagnant sections can extend replacement; validate recovery in the actual assembly and retain that test with the operating procedure.

Q4. Can I recalibrate the COD probe to remove restart spikes?

First investigate flushing, bubbles, optical fouling and timestamp alignment. Calibration cannot make stale or unrepresentative water equivalent to the current process sample.

Q5. How should unavailable data appear in the PLC or platform?

Use an explicit quality or availability state with a timestamp. Do not substitute zero COD or present a held last value as a new process observation. Test the downstream alarm behavior for this state.

Q6. What information helps specify a sampled COD station?

Provide the source location, tube dimensions, vessel volume, measured flow, pump schedule, expected COD/turbidity and pressure. Include the required time-to-decision so the sampling route and controller logic can be reviewed together.

Summary

COD sensor sampling needs proof that representative water reaches the instrument. Treat sample loss and restart as distinct operating states, validate the flush rather than guessing a fixed delay, and keep other water-quality channels independent. A salinity probe or any other responding instrument is not a substitute for checking the COD sample path.