Smart Coagulation: The Sensor‑Driven Playbook Municipal Plants Are Using to Lock In Low Turbidity

Tropical rivers can swing from ~20 NTU to >1000 NTU after rains, yet finished water targets keep tightening. Many plants aim for ≤0.3 NTU—often <0.1 NTU—in finished water (ypcadd.com), while Australian guidance puts filter effluent at <0.2–0.5 NTU (guidelines.nhmrc.gov.au). Indonesia permits up to 25 NTU in distribution (researchgate.net), yet utilities routinely operate well below that to safeguard disinfection and aesthetics (guidelines.nhmrc.gov.au; ypcadd.com).

The only way to keep pace with that variability is precise control of coagulation and flocculation. These core steps neutralize particle charges and aggregate colloids and natural organic matter (NOM) into settleable flocs, enabling low turbidity and pathogen removal (ypcadd.com; guidelines.nhmrc.gov.au). Operators lean on alum, ferric salts, and polyaluminum chlorides; in many plants, polyaluminum chloride (PAC) and aluminum chlorohydrate (ACH) are part of that toolkit, often sourced alongside general coagulants and flocculants for dose optimization.

Upstream solids removal still matters. Clarification and settling hardware pair with these chemistry moves, whether a conventional clarifier, a compact lamella settler, or a tube settler retrofit to boost capacity. Downstream, dual‑media filtration—such as beds built with sand/silica media or anthracite media—protects the finished‑water target and disinfection stage.

Regulatory targets and quality context

Indonesia’s Ministry of Health (Permenkes 492/2010) lists turbidity among required parameters (nuwsp.web.id). While turbidity is classified as a “parameter not directly related to health” with a maximum of 25 NTU (researchgate.net), day‑to‑day practice is stricter: U.S. utilities run at ≤0.3 NTU, many at <0.1 NTU (ypcadd.com), and Australian guidance targets filter effluent turbidity of <0.2–0.5 NTU and demands low turbidity at disinfection to ensure performance (guidelines.nhmrc.gov.au).

Real‑world Indonesian performance shows the baseline: an integrated coagulation–sedimentation–filtration train achieved an 85.8% turbidity reduction and met national standards (researchgate.net). The aim of modern control is to hold that efficiency under volatile raw‑water conditions.

Online sensors and field measurements

Automation starts with instrumentation. Nephelometric turbidity sensors (NTU, nephelometric turbidity units) before/after coagulation and filtration provide direct performance feedback, typically via 4–20 mA outputs to the control system. Online pH probes track pH shifts from dosing and their impact on floc formation.

Streaming‑current or zeta‑potential monitors measure the net electrostatic charge in conditioned water; targeting zero or a slight positive residual charge infers an optimum coagulant dose and enables feedback control (mdpi.com). UV254 (ultraviolet absorbance at 254 nm) or TOC/DOC (total/dissolved organic carbon) analyzers infer NOM load and coagulant demand for feed‑forward control. Flow meters anchor proportional dosing. Additional sensors—raw‑water temperature, conductivity, and turbidity—flag source‑quality shifts.

These instruments feed PLC/DCS/SCADA platforms—programmable logic controllers (PLCs), distributed control systems (DCS), and supervisory control and data acquisition (SCADA)—that adjust dose continuously. A field pilot in Indonesia wired a turbidity probe into a clarifier and, under sensor‑based control, took raw water at 109 NTU down to 20.54 NTU after coagulation (digilib.uinsgd.ac.id), illustrating the benefit of real‑time feedback (digilib.uinsgd.ac.id). Sensor taps are typically located just downstream of rapid mix and ahead of the clarification stage to capture the dosed‑water condition.

Control strategies: feed‑forward and feedback

Feed‑forward control uses upstream data—raw turbidity, UV254, color, or DOC—to set dose before the water reaches the mixer. The simplest rule is proportional‑to‑flow dosing; more advanced schemes correlate sensor inputs to jar‑test dose curves to anticipate storm spikes and adjust pre‑emptively.

Feedback control closes the loop on the dosed water itself. Streaming‑current control trims dose up when residual charge goes negative and down when it goes positive, holding a setpoint near neutral (mdpi.com). A Korean pilot used streaming current (target −0.05) and pH (target 8.0) as inputs to a fuzzy‑logic controller and cut a raw turbidity of 110 NTU to <10 NTU after coagulation/sedimentation (mdpi.com).

Final‑water turbidity is valuable for verification but is a poor real‑time control signal in conventional clarifiers due to 30+ minute lag. Exceptions include expedited processes (for example, Actiflo) or membrane filtration, where lag is shorter. Plants integrating coagulation ahead of membranes often align with ultrafiltration pretreatment or complete membrane systems to stabilize downstream performance.

Data‑driven “soft sensors” are emerging. Using plant history and jar tests, multivariate or AI models (ANNs, fuzzy logic) estimate optimal dose from inputs such as flow, turbidity, and TOC. Reviews report chemical savings in the tens of percent, with field cases up to ~75% lower coagulant demand in technical‑scale trials combining online coagulation with membrane filtration (mdpi.com; mdpi.com). Full‑scale utilities in Taiwan and Korea using fuzzy‑logic feedback with pH and turbidity inputs also reported efficiency and cost gains (mdpi.com; mdpi.com).

SCADA/DCS implementation details

In a typical architecture, sensors stream to a PLC where PID (proportional‑integral‑derivative) control blocks or logic adjust actuators such as dosing pumps or variable‑speed agitators. Logic often blends feed‑forward (e.g., high raw turbidity triggers a dose boost) with feedback (e.g., streaming current deviates from setpoint) and raises alarms on high final turbidity for operator review. A historian logs trends for compliance and tuning.

Off‑the‑shelf IEC‑61131 PLC code or DCS function blocks can integrate multiple inputs. Some plants run automatic jar tests with online sensors—such as a turbidimeter on a settling jar—to validate dose curves. Mechanical skids and sensor mounts fall under general water‑treatment ancillaries for plant integration.

Local case studies show feasibility. A small‑scale Indonesian WTP used a Raspberry Pi‑based SCADA for the coagulation stage, reading a turbidity probe and controlling mini‑pumps; under SCADA control, the 109 NTU feed was consistently reduced to ~20 NTU, and all elements (sensor, pumps, SCADA) operated without failure (digilib.uinsgd.ac.id).

Digitalization continues. “Smart Water” programs favor HART‑enabled sensors with remote diagnostics and DCS/SCADA integration. IoT and cloud platforms are emerging for auxiliary monitoring such as tank levels and distribution turbidity, while on‑coagulant control remains on‑site. Recent studies on IoT‑SCADA point to ML‑based forecasting for dose prediction and maintenance planning down the line.

Measured outcomes and operating benefits

Chemical savings are the headline. Automated control has delivered 20–75% lower coagulant consumption in published examples, including a technical‑scale trial with ~75% alum savings when online coagulation preceded membranes (mdpi.com; mdpi.com). Consistent low turbidity follows: in the Korean fuzzy‑control pilot, raw 110 NTU was cut to <10 NTU after coagulation/sedimentation (mdpi.com), and many utilities require <1 NTU at disinfection to assure pathogen kill (guidelines.nhmrc.gov.au).

Filter performance and sludge handling improve with right‑sized dosing. Avoiding overdosing reduces aluminum carryover and extends filter runs, while less excess coagulant yields less sludge; for example, a 50% reduction in alum dose roughly halves alum hydroxide sludge volume.

Automation adds robustness, especially off‑hours. For unmanned nights or weekends, a streaming‑current controller can ramp dose with incoming storms, avoiding “anticipatory overdosing” that operators might otherwise apply (processprojects.net).

Design and tuning guidelines for plants

Sensor selection and placement: turbidimeters upstream and downstream; streaming‑current monitors after rapid mix and before settling; pH and UV sensors in the rapid‑mix train. All sensors require calibration (e.g., formazin standards for turbidity) and periodic cleaning to avoid drift.

Initial tuning via jar tests establishes dose curves across raw‑water scenarios, including the overdosing region to set safe bounds. Setpoints commonly include a streaming‑current target near neutral—often ±0.01–0.05 mA—and a settled‑water turbidity target, such as <5 NTU, for verification.

Control logic and tuning: PID or fuzzy‑logic loops are adjusted by stepping raw turbidity and observing response so the dose follows quickly with minimal oscillation, over the full seasonal range. Integration and redundancy: SCADA views trend turbidity, pH, and dose rate; alarms flag spikes or sensor faults; manual override and backup sensors are standard.

Maintenance: turbidimeters are checked against standards regularly (at least monthly), and streaming‑current probes need periodic cleaning. Performance logs help identify sensor bias. Operator training focuses on understanding the control logic and response plans; for example, when final‑water turbidity alarms, upstream sensors and a confirmatory jar test are checked.

Chemical choices and delivery are part of the envelope. Plants commonly deploy PAC or ACH as coagulants alongside alum and ferric salts, often sourced as PAC or ACH, and configured for precise feed via metering pumps. Specialty blends such as PAC/ACH are applied where raw‑water pH or NOM levels are challenging.

Examples, statistics, and compliance

Indonesian example: a conventional train (coagulation–flocculation–sedimentation–filtration) achieved 85.8% turbidity removal in Central Java and met Permenkes limits (researchgate.net), a baseline that automation can stabilize during storm events.

Industry statistics from improved operations indicate online coagulation control extends filter runs by 10–20% and cuts coagulant use by 20–50%. One utility reported a 25% chemical‑cost saving in the first year after installing streaming‑current‑based control.

Regulatory impact: keeping turbidity well below limits—e.g., below 5 NTU entering distribution, though up to 25 NTU is permitted by Indonesian law (researchgate.net)—supports compliance and defensible audits. Data logs from SCADA/historians provide verification records.

Bottom line for operators and control engineers

Automated control of coagulation/flocculation—anchored by turbidity, streaming‑current, pH, UV254, and flow measurements—stabilizes treatment, often lowers chemical use by tens of percent, and protects downstream processes and disinfection (ypcadd.com; guidelines.nhmrc.gov.au; mdpi.com). In Indonesia, where Permenkes and SNI compliance is the bar, upgrading to well‑tuned SCADA/PLC control with clear setpoints and routine calibration is a pragmatic path to resilient, lower‑cost operations (nuwsp.web.id; mdpi.com).