How a dosing pump is controlled determines whether the plant under-doses (fails consent), over-doses (wastes chemical) or holds setpoint reliably. This page covers feedback, feed-forward and cascade control modes; PID tuning for water-treatment dosing; sensor selection and placement; and SCADA integration.
The three building blocks of every dosing control loop.
Every chemical-dosing control system reduces to one of three modes, or a combination. Feedback measures the result and adjusts the dose to correct error — reliable but slow. Feed-forward measures the disturbance (flow, influent quality) and adjusts dose proactively — fast but uncalibrated. Cascade combines both: feed-forward provides the bulk dose, feedback trims for accuracy. Choosing the right combination depends on process dynamics, available sensors and consequence of error.
The classic PID loop: measure the output, correct the dose.
Downstream analyser (pH, ORP, chlorine, turbidity) measures the controlled variable. A PID controller calculates the error from setpoint and adjusts dosing pump speed or stroke to drive error to zero.
Strengths: robust to unmeasured disturbances; self-correcting; needs only one analyser.
Weaknesses: reactive — corrects after the process has drifted; performance limited by deadtime between dose injection and analyser response. Large deadtime (long pipe runs, slow reactions) makes pure feedback unstable or sluggish.
pH neutralisation (fast reaction, short loop, robust pH sensor). Free-chlorine residual hold (slow process, long contact tank). Wherever loop deadtime is short relative to acceptable error duration.
| Application | P (gain) | I (reset, min) | D (rate, s) |
|---|---|---|---|
| pH (single-stage) | 0.5–2 | 0.5–3 | 0 |
| pH (cascade lime + acid) | 0.2–1 | 1–5 | 0–5 |
| Free chlorine | 1–3 | 2–10 | 0 |
| ORP (oxidation) | 0.5–2 | 1–5 | 0 |
| Turbidity trim (coagulant) | 0.2–1 | 10–30 | 0 |
Always re-tune after commissioning; published gains are starting points only. Use auto-tune feature on modern PLCs as the first pass.
Look at the input, calculate the dose, deliver it — before the disturbance reaches the process.
Influent flow meter (or composition analyser) drives dosing pump speed via a calibrated function. Example: ratio-paced coagulant dosing where pump speed is directly proportional to plant flow. There is no feedback loop — the control assumes the calibration is accurate.
Strengths: fast response (no deadtime); inherently stable; simple to commission.
Weaknesses: no self-correction. If the calibration drifts (due to seasonal water quality changes, polymer batch variation, etc.), the process slowly drifts away from optimum.
Coagulant flow-pacing in drinking-water plants. Polymer flow-pacing in clarifiers. Chlorination of constant-quality groundwater. Wherever the process is dominated by one disturbance variable and there is no fast, reliable downstream analyser.
Linear: dose = k · flow (kg/h = k · m³/h). Determine k by jar testing on representative samples and verify quarterly.
Calibration valid only within tested range. Outside it, performance degrades. Use feedback trim (cascade) to extend operating range without recalibration.
Seasonal at minimum. Storm events, source-water changes, polymer batch changes all warrant recalibration.
The best of both worlds — fast bulk dose + slow accuracy trim.
Feed-forward provides the bulk dose proportional to flow. Feedback PID adds a small trim correction based on residual analyser. Either pump speed adjusts via two stacked inputs, or two pumps (large bulk pump + small trim pump) work together.
Strengths: fast response to flow change AND self-correcting drift; tolerant of calibration error; handles wider operating range than either mode alone.
Weaknesses: more complex to commission; requires two reliable signals (flow + analyser); needs PID tuning that respects both inputs.
Most modern industrial and municipal plants. Particularly: drinking-water coagulation (flow + turbidity), chlorination (flow + residual), pH control (flow + pH), polymer dewatering (sludge mass flow + cake solids).
Output = (kflow × flow) + PID(residual error). Bound the PID output to ~±30% of bulk dose to prevent integrator wind-up if the analyser fails.
If residual analyser fails → fall back to feed-forward only with last-known offset. Alarm operator. Do not allow uncontrolled overshoot.
Bad sensor = bad control. Online analyser performance is the most common limitation.
Glass-membrane probes — reliable, fast, well understood. Replace every 12–24 months; calibrate weekly. Inline retractable mount preferred for solids-laden streams.
Robust for monitoring oxidising/reducing conditions (oxidation, disinfection, dechlorination). Slower response than pH; needs reference electrolyte refill.
Surface-scatter (low NTU) or ratio-turbidimeter (high NTU). Auto-clean wipers essential for fouling-prone water. Replace optics annually.
Amperometric (membrane) or DPD-colorimetric reagent-based. Amperometric: faster, cheaper; needs flow cell. DPD: more accurate at low residual; reagent Operating expenditure.
Ion-selective electrode (fast, Operating expenditure-low) or wet-chemistry analyser (accurate, slower). Critical for breakpoint chlorination control.
Inductive probes for high-fouling waters; contacting probes for clean water. Useful surrogate for alkalinity dosing or salt-load monitoring.
Long pipe runs between dose and analyser cause oscillation in pure feedback. Mitigation: shorten loop with closer analyser, or switch to feed-forward dominant.
If pump saturates at max speed but error persists, PID integrator keeps accumulating. When error clears, controller overshoots. Mitigation: clamping I-term, anti-windup logic in PLC.
pH electrode drift, fouled chlorine cell, vibration noise all corrupt the control signal. Mitigation: scheduled calibration; signal filtering; cross-checks with redundant sensors.
Operator manually adjusts setpoint during transients, hiding the real problem. Mitigation: clear alarms, trending dashboards, daily setpoint discipline; periodic loop audits.
Distributing local PID loops within a coordinated plant-wide control architecture.
Modern plants run dosing PID loops at PLC level for fast, deterministic response. SCADA above provides:
Industry 4.0 add-ons include dosing-pump health monitoring, predictive maintenance, dose-vs-result analytics and remote tuning via secure VPN.
Explore SCADA integrationDeterministic PID loops — 100–500ms scan. Pump speed control, residual measurement, interlocks.
Setpoints, recipes, alarming, trending, operator visualisation. 1–10 second update.
Multi-plant comparison, predictive analytics, chemical Operating expenditure dashboard, supplier integration. Daily / hourly.
PLC, SCADA and cloud architecture for water treatment plants.
Read MorePump type and accuracy considerations for each control mode.
Read MoreDocumentation of control loops, instrumentation and dosing-system topology.
Read MoreA common cascade-control application — flow-paced bulk + turbidity trim.
Read MoreSend us your loop performance data, P&ID and current PID gains. We will audit the control strategy, suggest mode upgrades (e.g., feed-forward addition), retune the PID and document SCADA integration improvements.
Our expertise spans multiple industries with sector-specific water treatment solutions.