Definition
Measured downstream distribution system water quality impact of AI-initiated process changes over the following 24 to 72 hours, catching adjustments that keep plant effluent compliant while degrading water in the network.
The Measurement Gap
A drinking water plant's AI optimizer adjusts coagulant dose based on real-time source water turbidity, reducing chemical costs by 14% over the first year. Plant effluent meets every water quality standard at the discharge point: turbidity, chlorine residual, pH, all well within limits. The compliance report is clean.
Three days after a significant coagulant dose reduction triggered by an AI-detected improvement in source water quality, a cluster of customer complaints arrives from a neighborhood 12 miles from the plant. Residents report discolored water and a metallic taste. Investigation reveals that the reduced coagulant dose produced treated water with slightly different chemistry (still compliant at the plant) that disturbed sediment deposits in aging cast iron distribution mains. The plant effluent was fine. The water arriving at customer taps was not.
This disconnect (between plant-level compliance and customer-level water quality) is invisible to every standard water treatment metric. Effluent monitoring tells you what left the plant. Distribution system monitoring (where it exists) tells you what arrived at specific monitoring points. Nobody connects AI process changes at the plant to water quality consequences in the distribution system with enough temporal specificity to identify which process change caused which downstream effect.
CLDIS fills this gap by correlating each AI-initiated treatment process change with downstream water quality measurements over the following 24-72 hours (adjusted for system travel time), weighted by the severity of the affected parameter. A rising CLDIS means the AI's process decisions are producing downstream consequences that plant-level monitoring cannot see.
Formula
Layer 1: Plain Language
For each AI-initiated process change (dosing adjustment, pH target change, disinfectant dose modification), measure the water quality at downstream distribution system monitoring points over the following 24-72 hours. Compute the change from the pre-adjustment baseline. Flag cases where plant effluent remained compliant but distribution system quality deteriorated.
Layer 2: Mathematical Notation
CLDIS = Σ (ΔWQ_distribution_i × w_i) / N_process_changes
Where:
- ΔWQ_distribution_i = maximum deviation in water quality at any downstream monitoring point within 72 hours of process change i, expressed as percentage of baseline
- w_i = severity weight based on the parameter affected (lead/copper: w=3.0, DBPs: w=2.5, chlorine residual loss: w=2.0, turbidity: w=1.5, taste/odor: w=1.0)
- N_process_changes = number of AI-initiated process changes in the measurement period
Units: Weighted percentage deviation per process change. Zero means no downstream impact. Positive values indicate downstream degradation.
Proposed formula, pending validation pilots.