Algae removal at Morton Jaffray, Harare

The clearest published demonstration of why flotation beats settling on African reservoir water — and an unusually honest account of what it costs you elsewhere in the treatment train.

Direct answer

A 60-litre dissolved air flotation pilot treating raw water from Lake Chivero removed 95% of chlorophyll-a and 64% of turbidity, against 52% and 53% for the conventional coagulation–settling process it would replace at Morton Jaffray Water Treatment Works. Flotation roughly doubled algae removal on the same water, with the same coagulant, in a parallel study run at the same time.

The study is equally clear about the cost: treated pH fell to 4.8–5.1 and electrical conductivity rose 42%, both consequences of the acid and alum regime rather than of flotation itself, and both requiring correction downstream.

95%Chlorophyll-a removal
64%Turbidity removal
7 m/hHydraulic loading rate
70 mg/LAlum dose after pH correction
614,000m³/day works capacity

The problem

Morton Jaffray Water Treatment Works supplies Harare from Lake Chivero and Lake Manyame, with a nameplate capacity of 614,000 m³/day. Its process train is conventional: coagulation and flocculation, sedimentation, filtration, disinfection and lime stabilisation.

Lake Chivero has been reported as eutrophic since the 1960s and has deteriorated since. The consequence for the works is the classic one: many algal species regulate their own buoyancy and will not settle, so a sedimentation basin cannot remove them. The published record for the plant describes filter backwash intervals of 4–8 hours against a design expectation of 24–36 hours, with the associated loss of plant output, and algae detected in the distribution system itself.

This is the exact situation dissolved air flotation exists for. A particle that will not sink because it is buoyant is a particle that floats easily once a bubble is attached to it. The comparison between flotation and settling turns almost entirely on particle density, and algae is the textbook case.

What was tested

ParameterValue usedLiterature range cited
Working volume60 L—
Hydraulic loading rate7 m/h5–15 m/h
Saturator pressure300–400 kPa300–600 kPa
Recycle ratio6–12%5–12%
Rapid mix, coagulant reaction5 min—
Flocculation2 min—
Retention time in flotation≈ 5 min5–15 min
Bubble generation1 mm nozzles from an elevated saturator—
CoagulantAlum at 70 mg/LTypical plant doses 15–50 mg/L
pH correctionSulphuric acid to pH 6.5–6.9Alum optimum 6–7
Campaigns7 experiments over 5 weeks, 30 min each at steady state—
Design and operating parameters as reported. The flotation stage was compared against a parallel conventional pilot — coagulation, flocculation and sedimentation — run in the same laboratory, on the same raw water, over the same period, which is what makes the comparison usable.

Two details are worth flagging for anyone reading this as a design precedent. The flocculation stage was two minutes, well below the 10–20 minutes normally specified for a full-scale flotation plant, and the loading rate of 7 m/h sits at the conservative end of the conventional 5–15 m/h band. A full-scale installation with proper flocculation would be expected to do better, not worse.

Raw water

ParameterRangeMeanCoefficient of variation
Turbidity3.14–4.04 NTU3.60 NTU10.5%
pH7.2–8.47.735.6%
Electrical conductivity328–483 µS/cm377 µS/cm19.3%
Chlorophyll-a0.36–2.12 µg/L1.31 µg/L68.1%
Seven composite samples, April to May 2014. Note the turbidity: at 3.6 NTU this is a low-turbidity water, which is precisely the condition under which settling performs worst and flotation performs best. Low mineral turbidity means light, low-density floc.

The alum requirement is the number that should catch a plant manager’s attention. At 70 mg/L after acid correction, Lake Chivero water needs well above the 15–50 mg/L typical of most treatment works — a direct cost consequence of the eutrophication, and one that compounds every year the lake deteriorates.

Results

ParameterTreated by DAFDAF changeConventional settling, same water
Turbidity1.01–2.02 NTU, mean 1.31−64%−53%
Chlorophyll-a0–0.36 µg/L, mean 0.07−95%−52%
pH4.8–5.1, mean 4.9−27% further−10%
Electrical conductivity—+42%+5%
All four changes were statistically significant at 95% confidence on a paired t-test. Turbidity and chlorophyll-a results are the case for flotation; pH and conductivity are the case against the acid-and-alum regime that was used with it.

The chlorophyll-a figure is the headline. Against the same raw water, with the same coagulant, in the same laboratory, in the same weeks, flotation removed 95% of the algal biomass indicator and settling removed 52%. That is not a marginal improvement; it is the difference between a filter that runs for a shift and a filter that runs for a day.

The caveats the study reports

  • Treated pH of 4.8–5.1 is outside any drinking-water guideline and must be corrected before distribution. Morton Jaffray already stabilises with lime, so the correction exists — but the lime demand would rise. This is a consequence of dosing sulphuric acid to reach the alum window and then dosing 70 mg/L of alum, both of which consume alkalinity. A coagulant with a wider pH window would reduce or remove the acid dose entirely; see PAC and ACH.
  • Conductivity rose 42%, against 5% for the conventional process, again driven by the acid and the alum rather than by flotation. It remained within the local potable standard.
  • It was a bench-scale pilot, not a full-scale installation. The results are a feasibility demonstration. A 614,000 m³/day works is a different proposition from a 60-litre perspex model, and the study says so.
  • Two-minute flocculation is short. This almost certainly understates what a properly designed full-scale flotation stage would achieve.
  • Raw chlorophyll-a was variable — a coefficient of variation of 68% — so the percentage removal carries real uncertainty around it.

What we would take from it

For any African works drawing on a eutrophic impoundment — and there are many — three things follow.

  1. Flotation is the correct process for algae, and the margin over settling is large. Roughly double the chlorophyll-a removal on identical water is not within the range of tuning a sedimentation basin.
  2. The coagulant choice deserves as much attention as the clarification process. Most of the downside reported here — collapsed pH, elevated conductivity, high chemical cost — comes from running alum at 70 mg/L behind an acid dose. A higher-basicity aluminium coagulant with a wider pH window would attack all three at once, and the study’s own recommendation is further work on the acid dosing regime.
  3. The economic case is in the filters, not just the clarifier. Backwash intervals of 4–8 hours instead of 24–36 hours mean a large fraction of the works’ capacity and washwater is being spent on algae that never should have reached the filters. That is where the payback sits.

Source

Feasibility of dissolved air flotation for drinking water treatment for Harare, Water Practice & Technology, volume 18, issue 1, page 102, IWA Publishing, 2023. Experimental work carried out at the University of Zimbabwe, April–May 2014, using raw water from the Morton Jaffray Water Treatment Works inlet. Published Open Access under CC BY 4.0. Read the full paper.

All figures on this page are as reported in that paper. Interpretation and the commentary under “what we would take from it” are ours and are not part of the published study.