A 2,500–10,000 m³/hr dissolved-air-flotation (DAF) and ultrafiltration (UF) pre-treatment facility lost stability when wet-season raw-water turbidity rose over 250 %. Using 275 logged operational records, we traced the cascade to three interdependent failures — and set out the fixes that restore resilience without major capital spend.
Treatment capacity across six parallel DAF streams.
Rise in intake turbidity entering the wet season (3.5 → 12.3 NTU mean, peak 146 NTU).
Logged operational entries analysed across dry and wet seasons.
Interdependent failures in saturator pressure, dosing and flocculation.
Anonymised from a six-month commissioning and early-operation dataset. Figures are the client’s logged operating data.
DAF followed by UF is a proven pre-treatment configuration for surface and reclaimed water — but the integrated process is sensitive to rapid swings in turbidity, organic loading and hydraulic shock. This facility treated a variable open surface-water source through six parallel DAF streams (ferric-chloride coagulation, acid pH correction, anionic-polymer flocculation) ahead of UF membranes. The transition from a calm dry season to a highly variable wet season was abrupt: mean intake turbidity tripled from 3.5 to 12.3 NTU, with isolated peaks above 140 NTU, while algae and organic loading rose — lifting coagulant demand and straining flocculation kinetics just as flows climbed toward 10,400 m³/hr.
Open surface-water source with seasonal turbidity excursions.
Ferric chloride (FeCl₃) dosing, target 2.5–4.0 mg/L as Fe.
Acid dosing to hold coagulation pH, typically 6.4–7.2.
Two-stage mechanical flocculation, variable-speed mixers (0–70 %).
Six parallel DAF streams — independent saturators, recycle pumps and scrapers.
UF feed tank receiving the combined DAF clarified water.
Streams 1 and 3 stayed consistent; Streams 2, 4 and 5 repeatedly exceeded 20 NTU. During several high-turbidity events outlet turbidity exceeded intake — negative removal, from sludge carryover, poor bubble–floc attachment or inadequate air dissolution.
| DAF Stream | Mean Outlet Turbidity (NTU) | Peak Outlet Turbidity (NTU) |
|---|---|---|
| Stream 1 | 5.4 | 27.1 |
| Stream 2 | 11.6 | 25.2 |
| Stream 3 | 6.8 | 21.7 |
| Stream 4 | 11.9 | 25.9 |
| Stream 5 | 7.4 | 33.6 |
| Stream 6 | 9.0 | 11.7 |
Target 4.2–5.5 bar, but logs recorded repeated drops to 1.5–3.0 bar — one recycle-pump fault ran ~10 h overnight. Open air-release valves and saturator blockages also featured.
A compressor failure stopped all coagulant and acid dosing until manually corrected; a closed ferric valve went undetected; acid-valve feedback faults caused intermittent over- and under-dosing.
Mixer trips during electrical storms lost first-stage flocculation on three streams; timed-scraper operation to save power caused sludge-hopper overflow under high solids.
DAF instability passed straight through to the membranes — raising fouling risk and threatening permeability.
| Parameter | Period A Mean (dry) | Period B Mean (wet) | Peak Recorded |
|---|---|---|---|
| UF Turbidity | 4.4 NTU | 10.5 NTU | 36.6 NTU |
| UF TSS | 3–8 mg/L | 5–37 mg/L | 37 mg/L |
| UF Feed pH | 6.86 | 7.20 | 8.00 |
On-site jar tests during wet-season operation gave floated-sludge TSS from 4 to 64 mg/L across identical intervals — inconsistent floc structure or bubble distribution even when upstream chemistry looked stable. pH tracked well at 6.8–7.2 but scattered badly above 7.5, pointing to acid-dosing capacity / mixing limits and sensor drift.
| Category | Share |
|---|---|
| Equipment faults / trips | 18 % |
| pH & control issues | 18 % |
| Chemical-dosing adjustments | 18 % |
| Sludge & scraper problems | 10 % |
| Weather-related events | 8 % |
| Pressure & flow anomalies | 6 % |
| Calibration & maintenance | 6 % |
The full operational dataset behind the findings, panel by panel.
Intake flow (blue) against turbidity (red, right axis). The sharp wet-season turbidity spikes — peaks above 140 NTU — against rising flow show the abrupt source deterioration the plant had to absorb.
Ferric, acid and polymer dosing over the period. Ferric was escalated through the wet season; acid dosing is intermittent, reflecting the valve and feedback faults that repeatedly disrupted coagulation.
Intake, dosed and UF-feed pH. They track well during stable periods but diverge during upsets — evidence of acid-dosing capacity / mixing limits and pH-sensor drift.
Outlet-turbidity distribution per DAF stream. Streams 1 and 3 stay tight and low; Streams 2, 4 and 5 spread high with frequent outliers above 20 NTU.
Turbidity, TSS and iron reaching the membranes. The wet-season excursions (turbidity to 36.6 NTU) map directly onto the upstream DAF instability.
Saturation pressure per stream. Repeated collapses from the 4.2–5.5 bar target down to 1.5–3.0 bar coincide with the poor-performing streams — the primary constraint on flotation.
Intake algae against UF turbidity. Elevated-algae periods cluster with higher UF turbidity, linking bloom loading to membrane-feed quality.
Logged operator comments per month, peaking in the wet-season months — a proxy for operational stress and intervention frequency.
Per-stream performance, dosing efficiency and control accuracy.
Removal efficiency by stream. Negative values (outlet worse than intake) are common on Streams 2, 4 and 5 — confirming sludge carryover and poor bubble–floc attachment; Streams 1 and 3 hold the highest median.
Ferric dose against residual iron in the UF feed. Higher dosing does not reliably raise residual iron, pointing to variable coagulation efficiency or ferric-hydroxide precipitation.
Dosed setpoint vs actual UF-feed pH against the 1:1 line. Points below the line indicate acid underdosing or mixing delay; the scatter above 7.5 shows control breaking down at high demand.
Online / offline / fault mentions per stream from the logs. Streams 2, 4 and 5 carry the most offline and fault mentions — consistent with their poorer turbidity removal.
The critical vulnerability was not the peak turbidity itself — it was the interdependence of three systems. When one failed, the others could not compensate.
The primary constraint. Low pressure cuts air-dissolution efficiency, producing larger bubbles and weaker floc–bubble agglomerates — directly raising outlet turbidity and sludge carryover.
A single air supply for dosing-valve actuation was a single point of failure; valve-position feedback faults broke closed-loop control of coagulant and acid.
No pre-emptive plan existed to raise coagulant, cut flow or switch scrapers to continuous before raw-water turbidity peaked.
Pressure monitoring with automatic alarms and standby-compressor switching to prevent pressure collapse; auto-reduce stream flow or take a stream offline below 3.5 bar.
Redundant air supply to dosing actuators with verified fail-safe positions, ending the single-point-of-failure on coagulant and acid delivery.
Link raw-water turbidity to automatic FeCl₃ dose escalation, removing operator lag during storm events.
Automatically switch scrapers to continuous operation when DAF inlet turbidity exceeds a threshold (e.g. 10 NTU) to stop hopper overflow.
Increase inline pH-sensor calibration frequency — drift was a recurring cause of dosing error.
A pre-emptive checklist for forecast heavy rain: FeCl₃ to 4.0 mg/L and polymer to 0.6 mg/L, cut flow 15–20 %, scrapers continuous, raise hopper-pump frequency.
Use Streams 1 and 3 as the baseline; investigate header configuration, orifice sizing and recycle ratio to lift the underperforming streams.
A full residence-time-distribution tracer study to check for short-circuiting/dead zones, and a sludge-volume index to predict carryover before it reaches the UF.
A DAF–UF pre-treatment plant that runs cleanly in the dry season can cascade into failure when the source deteriorates — not because of peak turbidity, but because saturator pressure, dosing reliability and flocculation stability are interdependent. Hardening those three, with automated storm protocols and redundancy, restores resilience without major capital. Detailed operational logging turned reactive fault-finding into predictive process management — and the facility now serves as a reference for exactly that.
Dissolved-air-flotation clarification engineered to the duty.
Read MoreMembrane filtration and pre-treatment for surface and reclaimed water.
Read MoreFeed-forward dosing, interlocks and alarms for resilient operation.
Read MoreMore projects across water and wastewater treatment.
Read MoreOur expertise spans multiple industries with sector-specific water treatment solutions.