Rolling out a PM2.5 sensor grid and turbidity/TOC monitoring on a 10-acre transfer station next month — how are you tying alerts to automatic misting and stormwater valve closures without piling up false positives? We’re running a compact baghouse, a small wet scrubber, and portable containment berms; looking for field-tested setpoints and calibration checks that kept your QA tight.
We’ve had best luck with a “2-of-3 + 5‑min persistence” rule before any actuation: two adjacent PM2.5 nodes >[redacted]/m³ for 5 min to start misting ([redacted]/m³ hysteresis to clear), and valves only close when turbidity >50 NTU AND TOC jumps >[redacted]/L over baseline for 10 min, with rain/wind-direction gating — think bouncer plus doorman. Add a daily auto‑zero on one PM node (HEPA cap) and a weekly 20 NTU check for turbidity; handy reference: https://www.epa.gov/air-research/air-sensor-guidebook — are you using rate‑of‑change suppression for spikes?
I’d gate misting on wind direction and humidity with a small rate-of-change test — only downwind spikes with dC/dt above your noise floor for about 3 minutes, plus a simple “RH>90% lockout”; escalate only when baghouse DP is about 15% over baseline… For stormwater, use a rolling median turbidity trigger that co-occurs with rainfall intensity before closing valves, and do daily zero + monthly formazin checks; EPA’s Air Sensor Toolbox has a solid primer: Air Sensor Toolbox | US EPA. It’s a bouncer for your PLC.
@claireb_87 One thing that cut our false closures: we only shut the stormwater valve when both turbidity and UV‑TOC sit >3× their rolling‑MAD above a 1‑hour baseline for about 7 min with rain rate >1 mm/hr, then we pulse‑close for 90 s and confirm the rise persists before committing; dawn UV drift can fool it, so we freeze the baseline near sunrise — do you have a tipping bucket on site?