Configure by Point, Not by Analyzer

COD monitoring in a drinking water plant is not a scaled-down version of wastewater compliance monitoring. The two serve different decision horizons. A wastewater plant needs an analyzable, defensible number at the outfall; a drinking water plant needs to know, before the water reaches coagulation or filtration, whether an organic load is arriving that the process cannot absorb. Starting with an instrument specification and then looking for a place to install it inverts this logic. The point should define the instrument, not the reverse.

At the raw water intake, the job is early warning, not adjudication. Storm runoff or seasonal turnover can lift organic load within hours, and the process window between intake and treatment is narrow. What this point requires is a continuous, low-maintenance signal that reports a direction of change—rising or stable—fast enough for pre-oxidant or coagulant dosing to respond. Reagent-free optical methods fit this function because they eliminate digestion delay and liquid waste, but their readings are optical equivalents tied to a site-specific calibration, not absolute referee values. Treating them as compliance instruments misreads their role.

Inside the process, a different question governs placement. A point after biological activated carbon, for example, exists to show whether that barrier is still removing organics, not to certify the final product. The instrument there should be chosen for sensitivity to the relevant removal mechanism and for stability under the local water matrix, not for the widest possible measuring range.

At the final outlet, the logic shifts again. If the data must be reported, compared, or defended, the method must be traceable to a standard reference—permanganate index or dichromate-based COD—even if that means accepting a slower or more maintenance-intensive instrument. Speed is no longer the binding constraint; comparability is.

This is the practical meaning of configuring by point: fast methods and reference methods are not competitors but layers of one system. The fast layer sees change; the reference layer confirms magnitude. Selecting a single analyzer type and distributing it uniformly across the plant collapses these layers into one, leaving either the intake blind to rapid shocks or the outlet unable to substantiate its own numbers. The point comes first because the decision the point exists to support comes first.