High-Fluoride Borehole Water Treatment: Iron Removal, Scaling Control and RO Concentrate Management
High-fluoride borehole water treatment has to be designed from the laboratory analysis, not from a standard membrane skid. A borehole in Tanzania’s Kilimanjaro border region, close to Kenya, produced water with 6.9 mg/L fluoride, 3.49 mg/L iron, 1,778 mg/L total dissolved solids (TDS) and very high bicarbonate alkalinity. Guangzhou Globaluf Water Treatment Co., Ltd. (GlobalUF) designed, factory-tested and shipped a 5 m3/h brackish-water reverse osmosis (RO) plant for this water. Iron removal, scaling control at 60% recovery, and a planned route for the fluoride-rich concentrate matter as much as the membranes.
This article can show the March 2024 borehole laboratory analysis, the process design, and a June 2026 factory water test in Guangzhou. That factory test used low-salinity municipal water, not the borehole. It cannot report site operating results, measured fluoride rejection on the well, or iron-removal performance in the field. The owner is installing and commissioning the plant on site. These design notes were originally published in Kenya Engineer on 15 November 2026: Treating High-Fluoride Borehole Water in East Africa: Inside a 5 m3/h Reverse Osmosis Plant.

What does the project video explain?
This GlobalUF video accompanies the design example. Factory test images and design calculations must not be interpreted as measured borehole treatment performance.
What does the March 2024 borehole analysis show, and which drinking-water limits apply?
The World Health Organization (WHO) guideline value for fluoride is 1.5 mg/L [1, 2]. The East African potable-water specification, adopted in Kenya as KS EAS 12:2018, sets the same fluoride limit, and limits iron to 0.3 mg/L, manganese to 0.1 mg/L and sodium to 200 mg/L [4]. WHO does not set a health-based guideline for TDS, but notes that water below about 600 mg/L is generally palatable and becomes increasingly unpalatable above about 1,000 mg/L [1].
Table 1 gives the March 2024 laboratory analysis of the borehole, reported against those limits. The main problems are high fluoride, high iron, high TDS and sodium, very high bicarbonate alkalinity, and turbidity, colour and bacteria. The pH of 7.48 is in the normal drinking-water range. The hardness of 176 mg/L as CaCO3 is only moderate, and is below the 300 mg/L KS EAS 12 limit for treated water [4]. This is not a high-pH or high-hardness water.
The laboratory listed both total alkalinity as 1,014 mg/L as CaCO3 and bicarbonates as 1,014 mg/L as HCO3−. Those two expressions are not chemically equivalent — the same alkalinity would be a higher figure if expressed as HCO3−. The scaling discussion below uses the reported alkalinity of 1,014 mg/L as CaCO3.
Groundwater along the East African Rift and the volcanic highlands may combine fluoride, iron, salinity and a heavy bacterial load. The Rift is one of the world’s well-documented high-fluoride belts [3]. Reverse osmosis can deal with most of these in one step, but only if pretreatment, scaling control and the concentrate are designed around the actual water.
| Parameter | Result | Limit |
|---|---|---|
| pH | 7.48 | 6.5–8.5 [1, 4] |
| TDS | 1,778 mg/L | palatability ~600–1,000 mg/L [1] |
| Electrical conductivity | 2,830 µS/cm | — |
| Turbidity / colour | 10 NTU / 64 Pt-Co | — |
| Iron / manganese | 3.49 / 0.06 mg/L | 0.3 / 0.1 mg/L [4] |
| Fluoride | 6.9 mg/L | 1.5 mg/L [1, 4] |
| Sodium | 500 mg/L | 200 mg/L [4] |
| Calcium / magnesium | 61 / 6 mg/L | — |
| Total hardness | 176 mg/L as CaCO3 | 300 mg/L, treated water [4] |
| Total alkalinity | 1,014 mg/L as CaCO3 | — |
| Bicarbonates | 1,014 mg/L as HCO3− | — |
| Sulphate / chloride | 110 / 50 mg/L | — |
| Silica | 50 mg/L as SiO2 | — |
| Nitrate | 9 mg/L as NO3− | — |
| Total coliforms, E. coli, total viable count | too numerous to count | not detectable / very low [1, 4] |
Source and scope: March 2024 borehole laboratory analysis, compared with WHO Guidelines for Drinking-water Quality (2022) [1, 2] and KS EAS 12:2018 [4], as also reported in the Kenya Engineer article (15 November 2026). A dash means a corresponding limit was not specified in the original laboratory comparison. The two 1,014 mg/L figures are not chemically equivalent; scaling calculations use alkalinity as CaCO3.
Why does this water scale at 60% recovery despite a normal pH and only moderate hardness?
At the design recovery of 60%, the concentration factor is 1 / (1 – 0.60) = 2.5. Calcium, bicarbonate, fluoride, sodium and silica in the concentrate are about two and a half times their feed values, and the concentrate pH typically rises (see Figure 2a below).
Calcium carbonate is the main mineral-scale risk because the bicarbonate alkalinity is very high for the amount of calcium present, not because the raw water is unusually hard or already high in pH. A standard Langelier calculation, using the reported pH, calcium and alkalinity at ordinary ambient temperature, already indicates a positive saturation index in the raw water. In the concentrate the index is higher still.
Silica would be about 125 mg/L in the concentrate. That figure is a design projection from the 50 mg/L as SiO2 feed result at 60% recovery, and should be checked in the same projection.
The plant has no acid dosing and no softener. Scaling is controlled by the 60% recovery and a scale inhibitor selected for this chemistry and dosed ahead of the RO. The choice should be confirmed with the supplier’s projection, rather than assumed from hardness alone. The pretreatment is not designed to remove a meaningful share of the hardness.
How is the 5 m3/h treatment train arranged, including iron removal?
Figure 1 and Table 2 show the treatment train as later adjusted. Raw water is collected in a 10 m3 owner-supplied tank — the same raw-water and reaction tank — aerated by a 5 m3/h, 0.75 kW blower. Sodium hydroxide (NaOH), polyaluminium chloride (PAC) and sodium hypochlorite (NaClO) are dosed by metering pumps through silicone hoses to the bottom of this tank. NaOH raises the pH, which favours oxidation of ferrous iron (the raw-water pH is already 7.48). PAC is the flocculant. NaClO is the oxidant and biocide. Bottom dosing plus aeration is intended to improve Fe2+ oxidation and to settle some turbidity. The original process drawing showed the flocculant and biocide at the iron-filter inlet; dosing was later moved to this tank for this reason. A 10 m3/h pump (35 m, 1.5 kW) then feeds the filters. For dosing arrangements on membrane plants more generally, see GlobalUF’s guide to membrane water treatment dosing systems.
Three media filters, each 1,000 mm in diameter by 2,100 mm and rated 8.5 m3/h, follow in series: a manganese sand filter with manganese sand (manganese dioxide) media, a sand filter, and an activated carbon filter. The manganese sand filter has to bring 3.49 mg/L iron down towards the 0.05 mg/L ferric-iron limit in typical membrane guidelines [6], and to take out the small amount of manganese. The sand filter lowers remaining particles, turbidity and the silt density index (SDI). The carbon’s main job is to remove residual chlorine from the NaClO before the polyamide RO membranes, which are damaged by oxidants. Manufacturers’ guidelines limit free chlorine at the membrane to 0.1 mg/L or less [6]. The carbon also helps with colour and organic matter. The same oxidation-and-filtration logic, for East African boreholes more generally, is discussed in How to Remove Iron from Borehole Water in Tanzania and East Africa. That article is a process guide; it is not a description of this plant.
After the carbon, a reducing agent and a scale inhibitor are dosed ahead of a 5 µm, 10 m3/h cartridge filter. The reducing agent is a backup to the carbon. The scale inhibitor provides the scaling control described above. Brackish-water RO at 60% recovery then produces 5.0 m3/h (feed about 8.3 m3/h). The high-pressure pump is rated 10 m3/h, 130 m (about 13 bar), 5.5 kW. Product water is disinfected with ultraviolet light and stored in a 5 m3 tank, also supplied by the owner. A clean-in-place skid (300 L tank; 10 m3/h, 1.5 kW pump) is included. Planning of membrane CIP cleaning triggers and methods is a separate operating task. There is no remineralisation or blending line on this plant.
| Item | Design duty or arrangement |
|---|---|
| Raw-water / reaction tank | 10 m3, owner-supplied; aerated; NaOH, PAC and NaClO dosed to the bottom |
| Blower | 5 m3/h, 0.75 kW |
| Raw-water pump | 10 m3/h, 35 m, 1.5 kW |
| Media filters (in series) | Three vessels, each 1,000 mm × 2,100 mm, rated 8.5 m3/h: manganese sand (manganese dioxide), sand, activated carbon |
| Cartridge filter | 5 µm, 10 m3/h, with reducing agent and scale inhibitor dosed ahead of it |
| RO section | Brackish-water RO at 60% recovery; product 5.0 m3/h; feed about 8.3 m3/h |
| High-pressure pump | 10 m3/h, 130 m (about 13 bar), 5.5 kW |
| Disinfection and product tank | Ultraviolet disinfection; 5 m3 product tank, owner-supplied |
| CIP skid | 300 L tank; 10 m3/h, 1.5 kW pump |
| Complete plant power | 400 V, three-phase, 50 Hz; total connected load pending confirmation |
| Not included on this plant | Acid dosing, softener, remineralisation, blending line |
Source and scope: Process design for this plant, as described in the Kenya Engineer article (15 November 2026). Duties are design ratings, not measured results from the borehole site. The 4 kW nameplate in Figure 4 must not be used to size the complete plant electrical supply.
What can a factory test on low-salinity Guangzhou water prove, and what can it not?
The 5 m3/h brackish-water RO skid was factory-tested in June 2026 in Guangzhou, before shipment (Figures 3 and 4). The water test used municipal, low-salinity factory water, not the borehole. The panel meters read about 259 µS/cm on the feed and 9.09 µS/cm on the permeate, an apparent salt rejection of about 96.5%. That shows the membrane elements and the instruments were working before the plant left the factory. It is not a measured result on the Kilimanjaro-region borehole, whose conductivity was 2,830 µS/cm and which also carries 6.9 mg/L fluoride and 3.49 mg/L iron.
A documented factory acceptance test can check assembly, interlocks, instrument response and hydraulic duties. The photographs alone do not verify the 5 m3/h design flow or 60% recovery. It cannot prove iron removal, fluoride rejection or antiscalant performance on the analysis in Table 1. Site commissioning should measure iron and turbidity at each filter outlet, SDI at the RO inlet, and product fluoride, TDS, sodium, iron and bacteria against WHO and KS EAS 12. Usual membrane feed limits are turbidity below 1 NTU, SDI below 5 (preferably below 3) and very low ferric iron [6]. Those checks are summarised in GlobalUF’s note on key parameters at RO feed water.


| Item | Factory test, June 2026, Guangzhou | What site commissioning should confirm |
|---|---|---|
| Test water | Municipal, low-salinity factory water | The Kilimanjaro-border borehole water (conductivity 2,830 µS/cm) |
| Feed / permeate conductivity | About 259 µS/cm / 9.09 µS/cm | Product TDS, fluoride, sodium, iron and bacteria against WHO and KS EAS 12 |
| Apparent salt rejection | About 96.5% on the factory water | Not a measured fluoride or salt rejection on the borehole |
| Hydraulics and build | Can support the 5 m3/h product flow, 60% recovery, interlocks and instrument response | Iron and turbidity at each filter outlet; SDI at the RO inlet |
| Iron removal, fluoride rejection, antiscalant performance on Table 1 | Cannot be proven by this factory test | To be measured during site commissioning |
Source and scope: June 2026 factory water-test readings and the March 2024 borehole analysis, as reported in the Kenya Engineer article (15 November 2026). Membrane feed limits follow DuPont Water Solutions, FilmTec™ Reverse Osmosis Membranes Technical Manual, Form No. 45-D01504-en [6]. The factory readings are for Guangzhou municipal / low-salinity factory water only.
How should the fluoride-rich concentrate be managed?
If fluoride rejection is high, almost all of the 6.9 mg/L in the feed reports to the concentrate. At 60% recovery that gives a concentrate fluoride of about 6.9 / 0.4 = 17 mg/L, and a brine flow of about 3.3 m3/h (Figure 2a). Those figures are design estimates, not measured site results.
In Kenya, discharge to the environment is regulated by the National Environment Management Authority under the Water Quality Regulations, which include a fluoride limit of 1.5 mg/L [5]. On the Tanzanian side of the border the concentrate also needs an approved route. Either way, 17 mg/L fluoride rules out uncontrolled discharge to surface water or a simple soakaway. An agreed route, such as a lined evaporation pond, should be part of the design.
The expected product fluoride depends on membrane rejection at this pH and temperature, and must be confirmed with the manufacturer’s projection and, later, by site tests. Brackish-water polyamide membranes often reject fluoride strongly enough at this pH to meet 1.5 mg/L from a 6.9 mg/L feed. That is a design expectation only, not a measured factory-test result. Remineralisation (for example a calcite contactor) or blending in pretreated raw water can be added at site if wanted [7, 8]. This plant has neither line. With raw water at 6.9 mg/L fluoride, blending is tightly limited (Figure 2b).

What power, shutdown and operating practices does the design assume?
The design supply is 400 V, three-phase, 50 Hz. Confirm total connected load and starting current against the final electrical schedule before sizing the supply. Most of the energy is the high-pressure pump. A simple estimate for that pump, at 13 bar, 70% efficiency and 60% recovery, is about 0.9 kWh/m3 of product, before the blower, raw-water pump and controls. That 0.9 kWh/m3 figure is a design estimate, not a measured site or factory-test result, and it is not the energy use of the complete plant.
Every shutdown, including a power failure, should end with a low-pressure flush, preferably with permeate [6, 9]. Daily logs of pressures, flows and conductivity, regular backwashing of the media filters, and local stocks of chemicals, cartridges and spares, matter as much as the original design. At least two people on site should be trained. Before the equipment is ordered, the design also needs a full analysis, a projection that checks calcium carbonate and silica at the design recovery, a defined iron-removal and SDI plan, oxidant removal before the membranes, an approved concentrate route, and trained operators with local spares.
Which published sources support these design notes?
- World Health Organization. 2022. Guidelines for Drinking-water Quality: Fourth Edition Incorporating the First and Second Addenda. Geneva: WHO. (Fluoride guideline value 1.5 mg/L, Table 8.9; TDS acceptability, Chapter 10.)
- World Health Organization. 2022. “Fluoride.” Chemical fact sheet, in Guidelines for Drinking-water Quality: Fourth Edition Incorporating the First and Second Addenda. Geneva: WHO.
- Fawell, J., K. Bailey, J. Chilton, E. Dahi, L. Fewtrell, and Y. Magara. 2006. Fluoride in Drinking-water. London: IWA Publishing, on behalf of the World Health Organization.
- Kenya Bureau of Standards. 2018. KS EAS 12:2018 Potable Water — Specification. Nairobi: KEBS.
- Republic of Kenya. 2024. Environmental Management and Co-ordination (Water Quality) Regulations, Legal Notice No. 177 of 2024, Third Schedule. Confirm the applicable discharge category and permit conditions with NEMA; Tanzanian requirements must be confirmed separately.
- DuPont Water Solutions. FilmTec™ Reverse Osmosis Membranes Technical Manual, Form No. 45-D01504-en. Wilmington, DE: DuPont. (Feed-water guidelines, Table 15; system shutdown and flushing.)
- Birnhack, L., N. Voutchkov, and O. Lahav. 2011. “Fundamental Chemistry and Engineering Aspects of Post-treatment Processes for Desalinated Water — A Review.” Desalination 273 (1): 6–22.
- World Health Organization. 2009. Calcium and Magnesium in Drinking-water: Public Health Significance. Geneva: WHO.
- DuPont Water Solutions. FilmTec™ RO/NF Systems Shutdown, Technical Manual Excerpt, Form No. 45-D01613-en. Wilmington, DE: DuPont.
Frequently asked questions about high-fluoride borehole water treatment
Can reverse osmosis remove fluoride from borehole water?
Brackish-water polyamide RO membranes often reject fluoride strongly enough, at a pH such as 7.48, to bring a 6.9 mg/L feed toward the 1.5 mg/L WHO and KS EAS 12 limit. That is a design expectation, not a measured result from the June 2026 Guangzhou factory test, which used low-salinity municipal water. It has to be confirmed with the membrane manufacturer and, later, by site tests.
What iron concentration is acceptable in RO feed water?
Typical membrane guidelines limit ferric iron in the RO feed to about 0.05 mg/L [6]. This borehole had 3.49 mg/L iron, so the design uses aeration, chemical dosing and a manganese sand filter (manganese dioxide media), then sand and carbon filters, before the membranes. Site commissioning should measure iron and turbidity at each filter outlet.
Is a factory acceptance test enough to prove fluoride rejection?
No. A factory test on low-salinity municipal water can prove the build, the interlocks, the hydraulic ratings and that the membranes and instruments respond on that water. It cannot prove fluoride rejection, iron removal or antiscalant performance on the borehole analysis in Table 1.
Can high-fluoride RO concentrate be discharged to a stream or a soakaway?
Not as an uncontrolled discharge, if concentrate fluoride is about 17 mg/L at 60% recovery. That 17 mg/L figure is a design estimate. In Kenya, environmental discharge is regulated by NEMA and includes a 1.5 mg/L fluoride limit [5]. On the Tanzanian side of the border the concentrate also needs an approved route, such as a lined evaporation pond agreed in the design.
Does this 5 m3/h plant blend or remineralise the product water?
No. This plant has no remineralisation or blending line. Blending in pretreated raw water is tightly limited when the raw water contains 6.9 mg/L fluoride. Those options can be added at site if wanted, but they are not part of the installed train.
What information should you send for a high-fluoride borehole RO review?
Send the following to [email protected] so the feed chemistry and the site constraints can be reviewed:
| Item | What to include |
|---|---|
| Feedwater analysis | A complete analysis, including fluoride, iron, manganese, TDS or conductivity, pH, hardness, alkalinity, silica, sodium, turbidity, colour and bacteria |
| Duty and water-quality target | Required product flow, intended use, and the target standard (for example WHO and KS EAS 12 potable limits) |
| Power and operating hours | Electrical supply (voltage, phases, frequency) and expected daily operating hours |
| Site responsibilities | The proposed concentrate disposal route, and whether tanks, civil works and installation will be supplied by the owner |
Source and scope: Review checklist based on the design issues in this article and on GlobalUF’s public note What Information Do You Need Before Requesting an RO System Quotation? The list is a starting point for a design review, not a complete tender specification.
GlobalUF supplies membrane water-treatment equipment, factory acceptance testing and engineering support. This is not an offer of turnkey civil construction. For equipment options, see the industrial and commercial RO systems page.
These design notes were originally published in Kenya Engineer on 15 November 2026: Treating High-Fluoride Borehole Water in East Africa: Inside a 5 m3/h Reverse Osmosis Plant.
Request a Quote
Fill out the form below to discuss your water treatment needs. Our experts will get back to you within 24 hours.
You may also be interested in:
