Water & Wastewater Treatment

The measurement list is fixed: alkalinity (EN ISO 9963), total and calcium hardness (ISO 6059), free CO₂, mineral acidity, suspended solids (EN 872), metal profile (EN ISO 11885) and temperature. A single grab sample represents none of them. A flow-proportional composite is taken over at least one full shift; on a surface source, wet and dry seasons are sampled separately. What sizes the plant is not the average but the peak-to-average ratio.

Streams are separated at source. If an acid pickling bath, a cyanide plating bath, a wet scrubber blowdown and general process water all merge in one pit, none of them can be treated at its own optimum pH. Blowdown from a flue gas treatment line, for instance, carries high chloride and sulphate; kept separate, its gypsum precipitation can be run under control, while blended it produces unexpected scaling right along the plant.

The output of characterisation is not a single table but a load profile: hourly flow, hourly acidity, hourly metal load. Equalisation volume is calculated from that profile. In practice 8–24 hours of equalisation halves the swing that dose control has to follow and gives a measurable drop in reagent consumption.

A treatment line usually carries several targets: discharge limits, receiving-water conditions, the pH tolerance of the biological stage, residual hardness and, for drinking water, the saturation index. Only one of them is binding, and that one governs the dose. If the binding target is a metal, the working pH follows the metal profile; if it is hardness, it follows the 10.3 or 10.8–11.3 band; if it is neutralisation alone, it follows 6.5–9.0.

Where that distinction is not made, the usual outcome is running at needlessly high pH. In a stream feeding a biological stage, going above pH 9 suppresses the nitrifiers and raises ammonium at the outlet, which then forces you to bring the lime-raised pH back down with acid. Choosing the right band from the start, instead of spending two chemicals in series, is the fastest saving available on most plants.

The titration curve comes first: lime slurry of known concentration is added in steps while pH is recorded. Flat plateaus on the curve identify the buffering items; the steep section shows the band where control will be difficult. In an effluent containing strong acid, the transition between pH 4 and 10 is almost vertical, and this curve shows on the design desk why a single-stage neutralisation basin will oscillate.

The jar test is run within the frame of ASTM D2035: six 1 L beakers, 1 minute of rapid mixing at 120 rpm, 15 minutes of slow mixing at 30 rpm, 30 minutes of settling. Each beaker is measured for pH, turbidity, 30-minute settled sludge volume and metal concentration. One detail is critical: metals must be measured both on a 0.45 µm filtered sample and on an unfiltered one. Discharge limits are written on total metal; when the filtered result looks perfect but the unfiltered one exceeds the limit, the problem is not chemistry but fine floc carryover, and the answer is a coagulant aid or filtration.

Sample age corrupts the result. A stored sample absorbs CO₂ from the air, ferrous iron oxidises and the true lime demand is measured wrongly; the test must be run within 24 hours. Moving from bench to plant, a scale-up factor of 1.1–1.3 is applied, because mixing uniformity and contact time in the real reactor sit below jar conditions.

There is an order in which a certificate is read: available CaO or Ca(OH)₂ content, loss on ignition (the carbonation indicator), grit content, fineness and, for quicklime, reactivity. Total CaO on its own is misleading; hard-burnt lime looks clean in analysis but slakes slowly, leaves grit and contributes nothing to the reaction despite appearing in the dose calculation. Drinking water duties add the heavy metal and impurity limits of EN 12518, with the test methods defined in EN 12485.

Incoming control is done on every consignment, not once a year. A two-point drop in available content translates directly into a pH shift on a fixed-volume dosing system and into a dose increase the operator cannot explain. Silo conditions matter as well: hydrated lime slowly carbonates with airborne moisture and CO₂, and a monthly loss of 1–3% in activity is normal in a silo whose vent air is not dried.

Slaking is the step that sets the quality of everything downstream. A paste slaker is run at a water-to-lime ratio of 2.5–3.5 : 1, a slurry slaker at 4–6 : 1. The slurry is aimed at 82–93 °C at the end of the reaction: with water too cold or too plentiful the lime is “drowned” and forms coarse, low-surface particles; taken to boiling, the particles agglomerate. Either way the reactivity of the Ca(OH)₂ obtained falls and a higher dose is needed for the same pH.

A grit classifier belongs at the slaker outlet. Varying between 2 and 10% with feed quality, this coarse abrasive fraction quickly eats pump rotors, valve seats and elbows if it is left in the line. Separated grit is also a quality signal: a sudden rise in the ratio is the first warning that the incoming lime is more hard-burnt than before.

The prepared slurry is held in a covered, continuously agitated tank. Solids begin to separate within 10–20 minutes of the agitator stopping, and a CaCO₃ crust forms on any surface in contact with air. Storage should not exceed 24–48 hours. Concentration is most practically tracked through density: a 5% slurry runs at about 1.03 g/cm³ and a 20% slurry at about 1.12 g/cm³, so a density meter on the ring main continuously verifies the number the dose calculation rests on.

The milk of lime line is built as a ring main that leaves the tank, passes the dosing points and returns to the tank, so that it is never stagnant. Velocity is held between 1.2 and 2.5 m/s: below that solids settle, above it erosion accelerates. Dead legs, long vertical drops and sharp elbows are where blockage starts; dosing valves are taken off the ring on short branches and flushed with clean water at the end of every shift.

Equipment selection follows the same logic. Hose (peristaltic) or progressive cavity pumps handle slurry well; diaphragm pumps with check valves block within weeks. On the valve side, pinch or knife gate valves are preferred. The injection point goes into a turbulent zone — right at the discharge of the mixer impeller — not onto a quiescent surface; otherwise a local zone above pH 12 forms, the metal hydroxides precipitated there never redissolve and a permanent crust builds at the pipe mouth.

In this environment the pH electrode is a consumable. CaCO₃ deposited on the glass slows the measurement and drifts it low over time, and the consequence is a system that overdoses lime. A retractable electrode holder, automatic jet cleaning and weekly two-point calibration (pH 7 and pH 10 buffers) remove that drift.

The reaction starts in the rapid mix zone: G value 300–1000 s⁻¹ for 1–3 minutes. Flocculation follows at G 20–80 s⁻¹ for 15–30 minutes, that is a Gt product in the 30,000–100,000 band. If those two stages are not separated, either nucleation stays insufficient or the floc formed is broken up by high shear and shows as fine turbidity in the clarifier.

Sludge recycle is the least known and most effective adjustment in this process. Returning part of the clarifier underflow to the reaction tank makes the existing CaCO₃ and metal hydroxide crystals act as seed; new solid grows on those seeds instead of nucleating as fines in solution. The result is a coarser, faster-settling and far better dewatering sludge — this is exactly the high-density sludge (HDS) loop in acid drainage treatment, which lifts sludge solids from 1–5% to 20–30%.

The control architecture cannot be single-stage. In a stream containing strong acid, the titration curve is nearly vertical between pH 4 and 10; trying to hold that band with one valve in one basin produces continuous oscillation. The right answer is two or three stages in series, each given a 2–4 pH unit duty, with a small-capacity trim valve on the last stage. Add feed-forward computed from inlet flow and acidity and the system responds to a load spike without waiting for the measurement lag.

On the sludge side lime does two jobs. For conditioning, lime at 10–25% of dry solids together with ferric chloride lowers specific filtration resistance and allows a filter press to deliver a cake at 25–35% dry solids, where a polymer-conditioned belt press stays at 18–22%. The effect is quantified by capillary suction time (EN 14701-1) or specific resistance in a Büchner funnel — judging by the look of the cake is misleading. For stabilisation, the target is to reach pH ≥ 12 and hold it for two hours, then stay above 11.5 for 22 hours. Two costs come with it: cake mass rises, and above pH 11 ammonium converts entirely to free ammonia, so the filtrate returns to the plant inlet with a high nitrogen load. Lime-stabilised biosolids can, within the applicable limits, be used as a liming material in agriculture and soil improvement.

Seasonal adjustment is a natural part of this process and has two physical causes. First, calcium carbonate has retrograde solubility: it dissolves more readily in cold water. Holding the same residual hardness in winter therefore requires the pH to be lifted by 0.2–0.3 of a unit. Second, viscosity: water at 1.31 mPa·s at 10 °C falls to 0.89 mPa·s at 25 °C, so the same floc settles roughly a third more slowly in cold water. That is why winter calls for a lower surface loading, lamella modules brought into service, or a longer flocculation time. Add the dilution of alkalinity in the wet season, and a fixed dose either falls short or wastes reagent and produces sludge for nothing.

During commissioning the dose is raised in steps, and at each step pH, turbidity, sludge volume, cake solids and effluent metal are recorded together. The aim is not to find a single operating point but to see where the curve plateaus and to set alarm bands accordingly. Water characterisation, jar test protocol, lime specification and dosing architecture can be worked out for your specific plant with our technical team.