WATER TREATMENT PLANT

PHASE 1: Remove the Big Stuff

The river water arrives. It's brown. Sticks, leaves, dead fish, plastic bags, gravel, sand. You can't treat water that has a tree branch floating in it. Before any chemistry begins, you need mechanical removal — screens and gravity. Bar Screens — The First Line Raw water enters through steel bar screens — parallel bars spaced 6-25 mm apart. These catch everything large: branches, rags, plastic bottles. A mechanical rake travels up the screen face, scraping collected debris into a dumpster. Flow rate through screens: Q = C_d × A × √(2g × h_L) Where: ├── C_d = discharge coefficient (~0.7 for clean bars) ├── A = net open area between bars ├── h_L = head loss across the screen └── g = 9.81 m/s² For a plant handling 500 ML/day (5.79 m³/s), you need roughly 15 m² of screen area. Two screens in parallel — one operating, one backup. Always.

Grit Chambers — Letting Sand Fall Out After screening, water flows into grit chambers. The idea: slow the water down enough that heavy particles (sand, gravel, silt) sink to the bottom, but light organic matter stays suspended and flows through. This is where Stokes' law governs everything: v = (ρ_p - ρ_f) × g × d² / (18 × μ) Where: ├── v = terminal settling velocity (m/s) ├── ρ_p = particle density (sand ≈ 2,650 kg/m³) ├── ρ_f = fluid density (water ≈ 1,000 kg/m³) ├── d = particle diameter (m) ├── μ = dynamic viscosity of water (0.001 Pa·s at 20°C) └── g = 9.81 m/s² For a 0.2 mm sand grain: v = (2,650 - 1,000) × 9.81 × (0.0002)² / (18 × 0.001) v = 1,650 × 9.81 × 0.00000004 / 0.018 v = 0.000648 / 0.018 v = 0.036 m/s = 3.6 cm/s A sand grain sinks 3.6 cm every second. Design the chamber so water moves horizontally slower than particles sink vertically, and the grit collects on the floor.

What We've Removed So Far ├── Debris >6 mm: bar screens (branches, rags, bottles) ├── Sand/gravel >0.2 mm: grit chambers (Stokes' law settling) ├── Still remaining: clay, bacteria, viruses, dissolved chemicals Turbidity reduction so far: from 100 NTU to about 80 NTU. We've barely started.

DESIGN SPEC UPDATED: ├── Bar screens: 6-25 mm spacing, mechanical rake, removes gross solids ├── Grit chambers: horizontal flow, 60s detention, captures sand >0.2 mm ├── Stokes' law: v = (ρp - ρf)gd² / 18μ — settling velocity ∝ d² └── Turbidity reduced from 100 NTU to ~80 NTU (20% improvement)

PHASE 2: Make Dirt Clump

The water is still cloudy. The particles causing that cloudiness are colloidal — so small (1-1,000 nanometers) that Brownian motion keeps them suspended forever. Gravity alone can never remove them. A colloidal clay particle would take 2 years to settle 1 meter. You need to make small particles become big particles. The trick: break their defenses. Why Particles Won't Clump Naturally Colloidal particles in water carry a negative surface charge. They repel each other — like trying to push two north-pole magnets together. The charge comes from mineral crystal structure and adsorbed ions. The zeta potential measures this repulsion. Raw water particles: -30 to -50 mV. At these values, particles stay dispersed indefinitely. To make them clump, you must neutralize that charge.

Coagulation — Chemical Charge Neutralization Add aluminum sulfate (alum): Al₂(SO₄)₃ When alum dissolves in water: Al₂(SO₄)₃ → 2Al³⁺ + 3SO₄²⁻ The Al³⁺ ions hydrolyze: Al³⁺ + 3H₂O → Al(OH)₃↓ + 3H⁺ Each Al³⁺ has a +3 charge. It neutralizes the negative surface charge on colloids. The aluminum hydroxide precipitate also forms sticky "floc" particles that sweep up everything in their path. Dose: typically 20-80 mg/L of alum (depends on raw water quality)

Flocculation — Gentle Stirring to Build Snowflakes Coagulation neutralizes charges (fast mix, 1-2 minutes). Flocculation builds large, settleable flocs (slow mix, 20-30 minutes). The coagulated particles are tiny. You need them to collide and stick — growing from microns to millimeters. Slow paddle mixers stir the water gently. Too fast and you shear the flocs apart. Too slow and particles never meet. Velocity gradient G = 20-70 s⁻¹ (slow flocculation) Compare: rapid mix G = 300-1,000 s⁻¹ (coagulant dispersal) Camp number: G × t = 10,000 - 100,000 For G = 40 s⁻¹, t = 30 min (1,800 s): G×t = 72,000 ✓

DESIGN SPEC UPDATED: ├── Coagulant: alum Al₂(SO₄)₃ at 20-80 mg/L, neutralizes negative colloidal charge ├── Rapid mix: G = 300-1,000 s⁻¹ for 1-2 min (disperse coagulant) ├── Flocculation: G = 20-70 s⁻¹ for 20-30 min (grow floc particles) └── Jar test every shift — optimal dose varies with raw water quality

PHASE 3: Let It Settle

The water is now full of fluffy floc particles — millimeter-sized clumps of aluminum hydroxide, clay, bacteria, and organic matter. They're heavy enough to sink. You need a basin big enough to hold the water still for long enough that everything falls to the bottom. Four hours. Half a million cubic meters of water. Just sitting there. Settling. Sedimentation Basin Design The critical parameter is surface overflow rate (SOR) — how fast water rises through the basin. If water rises faster than particles settle, particles escape over the top. SOR = Q / A Where: ├── Q = flow rate (m³/day) ├── A = surface area of basin (m²) └── SOR = surface overflow rate (m³/m²/day = m/day) For conventional floc settling: SOR = 20-40 m/day For a 500 ML/day plant at SOR = 30 m/day: A = 500,000 / 30 = 16,667 m² That's four football fields of basin surface area.

Hazen's Settling Theory Thomas Hazen, 1904. His insight: settling efficiency depends on surface area, not depth. A shallow basin with large surface area removes more particles than a deep narrow one with the same volume. Why? A particle needs to travel from the water surface to the basin floor to be captured. In a shallow basin, that distance is short. The particle has less chance of being carried out by the flow before it reaches the bottom. Removal efficiency ∝ A / Q (surface area / flow) Removal efficiency ≠ f(depth) (depth doesn't matter!) This is counterintuitive — a 1-meter-deep basin works as well as a 5-meter-deep one, IF they have the same surface area. This led to lamella plate settlers — stacked inclined plates that multiply effective settling area in a compact footprint.

Sludge — The Concentrated Waste Everything that settles out collects as sludge on the basin floor. A 500 ML/day plant produces roughly 10-15 tonnes of wet sludge per day. Sludge composition: ├── Water: 96-99% (it's mostly water) ├── Aluminum hydroxide: 30-50% of dry solids ├── Clay and silt: 20-40% ├── Organic matter: 10-20% └── Bacteria and pathogens: concentrated 100-1,000× from raw water This sludge is scraped by mechanical collectors to a hopper, pumped out, thickened, dewatered, and sent to landfill. The sludge is not hazardous waste — it's mostly clay and aluminum salts — but it's not exactly pleasant. After sedimentation: turbidity drops from 80 NTU to 2-5 NTU. That's a 95-97% reduction. But 2 NTU still contains millions of particles per liter.

DESIGN SPEC UPDATED: ├── Surface overflow rate: 20-40 m/day for floc settling ├── Basin area: A = Q / SOR → 16,667 m² for 500 ML/d (four football fields) ├── Hazen theory: efficiency depends on surface area, not depth ├── Lamella plates: 6-10× effective area, 1/6th footprint └── Turbidity: 80 NTU → 2-5 NTU (95-97% removal)

PHASE 4: Filter the Rest

The water looks clear — 2 NTU is transparent to the eye. But "looks clean" isn't clean. At 2 NTU, there are still millions of particles per liter, including bacteria and Cryptosporidium cysts that resist chlorine. You need to push water through a physical barrier so fine that essentially nothing gets through. Rapid Sand Filtration — The Workhorse The standard: a dual-media filter bed:

How Filtration Actually Works Particles aren't just "sieved out" by the gaps between sand grains. The gaps are much larger than most particles. Instead, three mechanisms capture particles: ├── Straining: particle too big to fit between grains (minor role) ├── Adsorption: particle sticks to grain surface via van der Waals + electrostatic ├── Interception: particle follows streamline close enough to contact grain └── Sedimentation: particle settles onto grain within pore space The coagulant residual from Phase 2 is critical here. The remaining Al(OH)₃ coating on particles and grains creates "sticky" surfaces. Without prior coagulation, sand filtration removes only 50-70% of particles. With it: >99%.

Backwashing — Cleaning the Filter After 24-72 hours, the filter bed is clogged with captured particles. Head loss across the bed rises from 0.3 m to 2.5 m. Time to clean. Backwash process: ├── Step 1: close inlet, drain water above bed ├── Step 2: pump clean water UPWARD through the bed at 40-50 m/h ├── Step 3: bed fluidizes — grains lift and separate, trapped dirt released ├── Step 4: air scour simultaneously (bubbles agitate grains) ├── Step 5: dirty backwash water flows to waste (or recycled to plant inlet) ├── Step 6: settle bed for 5 min, return to service └── Total time: 15-20 minutes Backwash uses 2-5% of total plant production. A 500 ML/d plant sends 10-25 ML/day back through the filters just for cleaning. After filtration: turbidity drops from 2-5 NTU to <0.1 NTU. Pathogen removal: 2-3 log (99-99.9%).

DESIGN SPEC UPDATED: ├── Dual-media filter: 0.6 m sand + 0.3 m anthracite + 0.3 m gravel support ├── Filtration rate: 5-15 m/h, run time 24-72 hours ├── Capture mechanisms: adsorption + interception + sedimentation (not just sieving) ├── Backwash: 40-50 m/h upflow, 15-20 min, uses 2-5% of production └── Turbidity: 2-5 NTU → <0.1 NTU (99%+ particle removal)

PHASE 5: Kill Everything

The water is crystal clear. Turbidity below 0.1 NTU. But clarity doesn't mean safety. A single milliliter can still contain thousands of bacteria and hundreds of viruses. You can't see a bacterium — it's 1 micrometer, 500 times thinner than a human hair. You need to kill them. All of them. And keep the killing agent active all the way to the consumer's tap. Chlorination — The Chemistry of Killing Chlorine gas dissolves in water: Cl₂ + H₂O → HOCl + H⁺ + Cl⁻ The active killer is hypochlorous acid (HOCl). It's an oxidizer — it rips apart cell membranes, denatures proteins, and destroys DNA. Nothing biological survives prolonged contact. HOCl dissociates depending on pH: HOCl ⇌ H⁺ + OCl⁻ At pH 7.5: ~50% HOCl, ~50% OCl⁻ At pH 6.5: ~90% HOCl (much more effective) HOCl is 80-100× more effective than OCl⁻ at killing pathogens. This is why pH control matters enormously — the difference between pH 7 and pH 8 can mean a 3× increase in required chlorine dose.

The CT Concept — How Much Kill Is Enough Disinfection isn't instant. It takes TIME and CONCENTRATION. The product of both determines the kill: CT = C × T Where: ├── C = disinfectant concentration (mg/L) ├── T = contact time (minutes) └── CT = mg·min/L Required CT values for 99.99% kill (4-log) at pH 7, 20°C:

A 500 ML/d plant typically maintains: ├── Chlorine dose: 2-4 mg/L at the plant ├── Contact time: 30 minutes minimum (in a contact basin or baffled tank) ├── Residual at plant exit: 1.0-1.5 mg/L └── Residual at far tap: >0.2 mg/L (this is the legal minimum)

Disinfection Byproducts — The Dark Side of Chlorine Chlorine reacts with natural organic matter (NOM) in water to form: HOCl + NOM → THMs + HAAs ├── THMs (trihalomethanes): chloroform, bromoform — linked to bladder cancer ├── HAAs (haloacetic acids): linked to liver and kidney damage └── Regulatory limits: THMs <80 μg/L, HAAs <60 μg/L The paradox: chlorine saves millions from cholera and typhoid, but creates tiny cancer risks. The solution isn't removing chlorine — it's removing organic matter BEFORE adding chlorine. That's why Phase 2 (coagulation) and Phase 4 (filtration) matter so much.

DESIGN SPEC UPDATED: ├── Chlorination: Cl₂ + H₂O → HOCl (the active killer) ├── pH critical: HOCl 80-100× more effective than OCl⁻, operate at pH 7.0-7.5 ├── CT = C × T: Giardia 4-log requires CT=104 mg·min/L ├── Cryptosporidium: chlorine-resistant, must be removed by filtration + UV └── DBPs: THMs and HAAs from chlorine + organic matter, limit by pre-removing NOM

PHASE 6: The Invisible Threats

The water is clear, disinfected, and free of particles. But dissolved in it — invisible to any filter — are chemicals measured in parts per billion. Lead from old pipes. Arsenic from geology. PFAS from firefighting foam. You can't see them, smell them, or taste them. They cause cancer, neurological damage, and developmental harm. Over decades. Lead — The Pipe Problem Flint, Michigan. 2014. The city switched water sources but didn't add corrosion inhibitors. The new water's chemistry dissolved lead from old service lines. Blood lead levels in children doubled. Lead dissolves from pipes when water is corrosive: ├── Low pH (<7.0) → aggressive to lead ├── Low alkalinity → no buffering capacity ├── Low calcium → no protective scale layer ├── High chloride/sulfate ratio → accelerates corrosion The fix: add orthophosphate at 1-3 mg/L PO₄. It forms a lead phosphate mineral scale inside pipes that seals the lead away from the water. Takes 6-12 months to fully form.

Arsenic — The Geological Poison Arsenic occurs naturally in groundwater — dissolved from bedrock. Bangladesh: 77 million people exposed to arsenic >10 ppb from tube wells. Causes skin lesions, cancers, cardiovascular disease over 10-20 years. Removal methods: ├── Coagulation with ferric chloride: As adsorbs onto Fe(OH)₃ floc, removes to <5 ppb ├── Ion exchange: strong-base anion resin, selective for arsenate ├── Activated alumina: adsorption, effective but pH-sensitive └── Reverse osmosis: removes 95%+ (Phase 8) EPA maximum: 10 ppb. WHO guideline: 10 ppb. Some countries still allow 50 ppb.

PFAS — The Forever Chemicals PFAS (per- and polyfluoroalkyl substances) — used in non-stick pans, firefighting foam, waterproof clothing. The C-F bond is one of the strongest in organic chemistry: 485 kJ/mol. Nothing in nature breaks it. Hence "forever chemicals." Found in drinking water at ppt (parts per trillion) levels. Linked to cancer, thyroid disease, immune suppression. Removal: granular activated carbon (GAC) GAC surface area: 1,000 m²/g. One gram of activated carbon has the internal surface area of 4 tennis courts. PFAS molecules adsorb onto this enormous surface through hydrophobic interactions.

DESIGN SPEC UPDATED: ├── Lead: corrosion control with orthophosphate, Flint crisis from missing inhibitor ├── Arsenic: coagulation with FeCl₃ or ion exchange, EPA limit 10 ppb ├── PFAS: GAC adsorption, 1,000 m²/g surface area, replace every 6-18 months └── These threats are invisible — require analytical chemistry to detect at ppb/ppt levels

PHASE 7: Soften or Scale

Run hot water through your kettle for a year. See that white crusty buildup? That's calcium carbonate scale — dissolved calcium and magnesium that precipitate when heated. Now imagine that happening inside every hot water pipe, boiler, and heat exchanger in a city. Hard water costs industries billions in damaged equipment. But remove too much hardness and the water becomes corrosive, attacking the pipes themselves. What Makes Water "Hard" Hard water contains dissolved Ca²⁺ and Mg²⁺ ions, picked up from limestone and dolomite geology.

Lime-Soda Softening The classic approach: add lime (Ca(OH)₂) and soda ash (Na₂CO₃). These raise the pH, converting dissolved Ca²⁺ and Mg²⁺ into insoluble precipitates that settle out. For calcium removal: Ca²⁺ + Ca(OH)₂ + CO₂ → 2CaCO₃↓ The dissolved calcium becomes solid calcium carbonate — it literally falls out of the water as a white powder. For magnesium removal (requires higher pH): Mg²⁺ + Ca(OH)₂ → Mg(OH)₂↓ + Ca²⁺ Then the new Ca²⁺ is removed with soda ash: Ca²⁺ + Na₂CO₃ → CaCO₃↓ + 2Na⁺ Lime-soda softening reduces hardness to 80-120 mg/L. Good enough for most municipal supply. Going lower requires ion exchange.

Ion Exchange — The Molecular Swap An ion exchange resin is a bed of tiny plastic beads (0.5-1.0 mm) coated with charged sites. Each site holds a sodium ion. When hard water flows through, the resin swaps Na⁺ for Ca²⁺ and Mg²⁺: Resin-2Na⁺ + Ca²⁺ → Resin-Ca²⁺ + 2Na⁺ The calcium sticks to the resin. The sodium goes into the water. Hardness drops to <5 mg/L. When the resin is exhausted (all Na⁺ swapped for Ca²⁺): ├── Backwash with 10% NaCl brine ├── Mass action drives Ca²⁺ off the resin, Na⁺ back on ├── Regeneration time: 30-60 minutes ├── Brine waste: high-salinity discharge (environmental concern) └── Resin life: 5-10 years

DESIGN SPEC UPDATED: ├── Hard water: Ca²⁺ + Mg²⁺ >120 mg/L as CaCO₃ → scale in pipes and equipment ├── Lime-soda softening: precipitates Ca/Mg as CaCO₃ and Mg(OH)₂ → 80-120 mg/L ├── Ion exchange: resin swaps Na⁺ for Ca²⁺/Mg²⁺ → <5 mg/L, regenerate with NaCl └── Balance: too hard = scale damage, too soft = corrosion — target 80-120 mg/L

PHASE 8: Push Through Membranes

What if your source isn't a river but the ocean? Seawater: 35,000 mg/L dissolved salts. Drinking water limit: 500 mg/L. You need to remove 98.6% of everything dissolved in it. No chemical can do this. No filter bed is fine enough. You need to push water through a membrane with pores so small that only H₂O molecules fit through — and leave everything else behind. Osmotic Pressure — The Force You Must Overcome Put saltwater on one side of a semipermeable membrane and freshwater on the other. Water naturally flows from fresh to salt — osmosis. Nature tries to equalize concentration. The pressure needed to STOP this natural flow is the osmotic pressure: π = iMRT Where: ├── i = van 't Hoff factor (NaCl: i ≈ 2, because it splits into Na⁺ + Cl⁻) ├── M = molarity (seawater ≈ 0.6 M NaCl) ├── R = gas constant (0.0821 L·atm/mol·K) ├── T = temperature (298 K = 25°C) For seawater: π = 2 × 0.6 × 0.0821 × 298 π = 29.3 atm ≈ 27 atm To push water BACKWARDS through the membrane (reverse osmosis), you must exceed this pressure. Operating pressure for seawater RO: 55-70 atm (about 2× the osmotic pressure — the extra drives flow rate).

Energy Cost — The Real Limit of Desalination Pushing water through a membrane at 60 atm requires enormous energy: Theoretical minimum: 1.06 kWh/m³ (thermodynamic limit) Practical consumption: ├── Seawater RO (current best): 3.0-3.5 kWh/m³ ├── Seawater RO (average plant): 4-5 kWh/m³ ├── Brackish water RO: 0.5-2.5 kWh/m³ └── Conventional treatment: 0.2-0.5 kWh/m³ Desalination uses 10-25× more energy than conventional treatment. For a 500 ML/d seawater RO plant at 4 kWh/m³: Daily energy = 500,000 m³ × 4 kWh/m³ = 2,000,000 kWh/day That's the output of a 100 MW power plant running 20 hours a day — just for water. This is why desalination remains a last resort, used mainly in water-scarce regions (Middle East, Singapore, parts of Australia).

Membrane Fouling — The Eternal Enemy RO membranes clog. Biological growth, mineral scaling, and colloidal fouling reduce flux and increase energy consumption. Prevention: ├── Pre-filtration to <5 μm (ultrafiltration pre-treatment) ├── Anti-scalant chemicals (prevent CaCO₃ and CaSO₄ precipitation) ├── Periodic chemical cleaning (acid wash, caustic wash) ├── Membrane replacement: every 5-7 years, cost $500-1,500 per element └── A large plant has 10,000-50,000 membrane elements Recovery rate: 40-50% for seawater (you get 400-500L fresh per 1,000L feed). The other 500-600L is concentrated brine — twice the salinity of seawater — discharged back to the ocean through diffusers.

DESIGN SPEC UPDATED: ├── Osmotic pressure: π = iMRT → seawater π ≈ 27 atm ├── RO operating pressure: 55-70 atm (2× osmotic pressure) ├── Energy: 3-5 kWh/m³ (10-25× conventional treatment) ├── Recovery: 40-50% for seawater, rest is brine discharge └── Membrane pores: ~0.1 nm — molecular-level separation

PHASE 9: Distribute Without Recontamination

You've built the perfect treatment plant. The water leaving it is pristine. Now you need to move it through 5,000 km of buried pipes — some over 100 years old — to 500,000 taps. The pipes are dark, warm, and wet. The perfect environment for biofilm growth. Every dead-end pipe, every service line, every water main break is a chance for recontamination. Pipe Flow — The Hazen-Williams Equation Water doesn't flow for free. Pipe friction consumes pressure. The Hazen-Williams equation predicts flow: v = 0.849 × C × R^0.63 × S^0.54 Where: ├── v = flow velocity (m/s) ├── C = roughness coefficient (new pipe: 140, 50-yr old pipe: 80) ├── R = hydraulic radius (pipe diameter / 4 for full pipe) ├── S = hydraulic gradient (head loss / pipe length) The roughness coefficient C drops with age. A new ductile iron pipe (C=140) carries water with low friction. After 50 years of tuberculation (rust nodules), C drops to 80. Flow capacity decreases by ~40% without any change in pipe diameter.

Chlorine Residual Decay The chlorine residual you send out of the plant decays as it travels through the network. It reacts with: ├── Pipe walls (corrosion products consume chlorine) ├── Biofilm on pipe surfaces (bacteria consume chlorine) ├── Residual organic matter in the water └── Sediment in dead-end mains Decay follows first-order kinetics: C(t) = C₀ × e^(-kt) Where: ├── C₀ = initial chlorine (1.0-1.5 mg/L at plant exit) ├── k = decay constant (0.1-0.5 per hour, depends on pipe condition) ├── t = time in system (hours) For a customer 24 hours from the plant, with k = 0.15/hr: C(24) = 1.2 × e^(-0.15 × 24) C(24) = 1.2 × e^(-3.6) C(24) = 1.2 × 0.027 C(24) = 0.033 mg/L Below the 0.2 mg/L minimum. This customer is unprotected. Solutions: ├── Booster chlorination stations within the network ├── Reduce water age (flush dead ends, loop dead-end mains) └── Chloramination: use chloramine (NH₂Cl) instead — decays 10× slower

Dead Ends and Water Age The worst contamination risks are at dead-end mains — pipes that terminate with no through-flow. Water sits stagnant for days to weeks. Chlorine decays to zero. Biofilm blooms. Sediment accumulates.

DESIGN SPEC UPDATED: ├── Hazen-Williams: v = 0.849 × C × R^0.63 × S^0.54 — C drops with pipe age ├── Chlorine decay: C(t) = C₀ × e^(-kt) — first-order, k = 0.1-0.5/hr ├── Dead ends: stagnant water, zero chlorine, high bacteria — loop or flush └── Network: 5,000 km pipe, 500,000 connections, 24/7 pressure maintenance

PHASE 10: Monitor or Poison