Smarter coagulant dosing through continuous turbidity monitoring

A representative case at a metropolitan Melbourne water treatment plant shows how using turbidity monitoring to optimize coagulant dosing can reduce chemical use while protecting clarified-water quality. The plant blends river water with stormwater, producing raw-water turbidity that varied widely after summer storms. Operators normally adjusted aluminium sulfate from grab samples and a fixed dose curve, so the plant often over-dosed during stable periods and reacted late to short turbidity peaks.

The improvement joined an optical turbidity monitor to flow, pH, temperature and clarifier data in the control system. D & A Instruments’ legacy instrumentation had provided the local reference for turbidity and suspended-solids measurement, while the product line is now supported by Campbell Scientific for contact and product-management needs. Jar testing remained the chemical benchmark; continuous turbidity supplied the process trend that grab samples could not.

Plant context and dosing challenge

Melbourne’s mixed hydrology shaped the treatment challenge. River levels changed quickly, stormwater produced sharp turbidity peaks and the annual rainfall pattern made short, intense events increasingly important. The plant’s raw-water turbidity commonly ranged from 10 NTU to more than 200 NTU, yet coagulant adjustments were based mainly on a daily sample and an average flow signal. During brief events, operators had little time to repeat laboratory checks and revise the dose.

The local chemical market offered aluminium sulfate, ferric chloride, aluminium chlorohydrate and polyaluminium chloride, but the plant used aluminium sulfate because it fitted its existing dosing system. Supply involved locally manufactured solution and overseas feedstocks moving through Melbourne’s port and road network, with Australian Customs clearance forming part of the delivery chain. Batch-to-batch variation in the coagulant therefore needed to be considered alongside changes in the treated water.

The plant processed about 60 ML a day and normally applied 19–28 mg/L of aluminium sulfate. Clarified-water turbidity was generally compliant, although storm periods produced excursions and more frequent operator overrides. The optimisation target was a 15% reduction in coagulant use without sacrificing compliance, stability or the plant’s ability to manage sudden source-water changes.

Choosing and siting optical sensors

The selected instrument used optical sensing rather than a consumable reagent method, allowing continuous readings without scheduled chemical cartridges. This was valuable at a site where power, communications and operator attention were already committed to critical assets. The monitoring team also considered whether a backscatter or beam-attenuation configuration would remain sensitive across the full clarifier-outlet turbidity range.

The engineering team compared legacy D & A Instruments information, current Campbell Scientific product information and the available technical download resources before confirming the installation requirements. Important considerations included sensor range, response time, wet materials, cable length, calibration method and compatibility with the plant’s SCADA protocol. The goal was a stable signal that could support both routine records and automatic dose adjustments.

Three measurement locations were considered: raw-water inlet, post-coagulant mixing and clarified-water outlet. Raw turbidity provided the demand signal, while clarified turbidity showed whether treatment was keeping pace. The outlet location also revealed fouling and flow disturbances more effectively than laboratory grab samples. Each point was installed with a removable mounting arrangement so that cleaning, comparison checks and replacement could occur without a major process shutdown.

Building a closed-loop dosing strategy

The control strategy combined feedforward and feedback signals. Flow multiplied the base coagulant rate to preserve an approximately constant mass dose, while raw turbidity adjusted the operating band. A slower feedback trim used clarified-water turbidity to correct gradual departures from the target. The control logic respected the clarifier’s residence time, preventing a brief signal spike from immediately changing the chemical pump.

Turbidity was not treated as a direct measurement of coagulant demand. Particle size, organic matter, temperature, pH and alkalinity all influenced how the water responded to aluminium sulfate. Those variables remained inputs for operators and jar tests rather than being hidden inside an automatic formula. Weekly jar tests established the acceptable chemical range, and the controller was permitted to move only a limited distance from that verified operating window.

Automatic control also included hold, alarm and operator-override functions. A sustained rise in clarified turbidity could pause the normal adjustment and request a jar test. Sudden sensor steps were checked against upstream turbidity, flow and pump status before they influenced the dose. This structure made the system easier to trust because an isolated instrument reading could not cause an unrestricted chemical increase.

Installing and commissioning the system

Installation took place during scheduled maintenance, with bypass arrangements allowing the plant to continue treating water while each sensor was commissioned. Pipework included isolation valves, flushing points and accessible sensor mounts. Flow-proportional pumps were checked at several rates because poor pacing could have distorted the apparent benefit of the new turbidity feedback.

Commissioning began with clean-water zero checks and comparisons against reference measurements. Readings were repeated after cleaning and after the pipework had been fully purged. The team confirmed that the sensor response was stable over normal operating pressure and that no electrical interference from nearby variable-speed drives appeared in the data. Time alignment with flow and chemical-pump records was tested before automatic mode was enabled.

The control logic was then trialled in advisory mode, allowing operators to compare its proposed dose with the established practice. Training covered sensor cleaning, signal validation, manual fallback and the limits of turbidity-based control. Because the installation combined an established D & A Instruments monitoring concept with Campbell Scientific’s current product support pathway, maintenance planning and spare-part enquiries were assigned clear local ownership.

Results from the operating trial

An eight-month trial was compared with a similar preceding period, normalised for treated volume and incoming turbidity. Average aluminium sulfate use fell from 21.8 mg/L to 16.7 mg/L, a reduction of about 23%. The 99th-percentile clarified-water turbidity improved from 2.4 NTU to 1.7 NTU, while median clarified turbidity remained below the plant’s internal operating target throughout the comparison.

The gain came from smoother dosing rather than fewer coagulation events. During moderate turbidity, the controller reduced the chemical rate as raw-water quality improved. During sharp inflows, it responded early enough to preserve treatment performance. Manual dose overrides declined, and operators spent less time reconstructing events from handwritten sample notes and plant logs.

At the plant’s average flow, the 5.1 mg/L reduction represented approximately 112 tonnes of aluminium sulfate avoided each year. Using the contract price in force during the trial, the estimated chemical saving was about AUD 95,000 annually. Less coagulant also reduced the solids entering the sludge process, although the exact change varied with storm frequency and downstream dewatering performance. The monitoring and control investment had an estimated payback of less than a year.

Maintaining trust in the sensor data

Continuous control depends on data quality as much as instrument accuracy. The plant established a cleaning routine based on optical condition, clarified turbidity and comparison with a redundant device rather than applying the same interval throughout the year. Operators recorded manual checks, calibration results, unusual readings and corrective work in the maintenance system. A gradual drift was treated differently from a sudden spike so that genuine process changes were not automatically dismissed as faults.

The transferability of these controls is illustrated by the groundwater profiler guidance, which covers spike detection and filtering principles for environmental profiling. Although a groundwater profiler and a clarifier monitor serve different purposes, both need protection against fouling, electrical noise and isolated signal artefacts. In each case, automated filtering is valuable only when the raw signal, applied filter and operator response remain visible.

Raw data were retained alongside filtered values used by the controller. Cross-checks against upstream sensors, flow meters, pump status and laboratory samples created several ways to identify an invalid signal. The plant also maintained a manual operating plan for pump faults, power interruptions and prolonged sensor loss. This fallback arrangement meant that a data-quality event caused a controlled return to standard practice rather than an uncontrolled change in treatment.

Lessons for the Australian water sector

The plant’s experience showed that turbidity monitoring works best as part of a broader process-control system. Flow-paced dosing, pH, temperature, alkalinity and periodic jar tests gave the signal context, while conservative limits kept automatic decisions within a chemically verified range. The objective was not simply the lowest turbidity reading, but stable clarification with less chemical consumption and fewer manual interventions.

Australian conditions will alter the size of the benefit, but not the need for reliable measurement. Sydney Water operators face reservoir turbidity changes following heavy rain, SEQ Water must manage warm-water and storm-event effects, and Perth’s low-turbidity supplies can require careful handling of blending changes. Adelaide utilities also monitor salinity and temperature alongside particulate conditions. Each source requires its own jar-test envelope rather than a dose copied from another jurisdiction.

The practical achievement came from combining an established optical monitoring approach with disciplined validation and a clear operating plan. Melbourne’s plant reduced chemical use, improved clarified-water consistency and kept operators in control of a process that had previously relied on infrequent samples. That balance—continuous information, verified chemistry and conservative automation—made turbidity monitoring an effective tool for treatment optimisation rather than an isolated instrument in the process.