Separation, drying, product quality testing, energy integration with MVR, plant layout, and process control/safety systems — completing the full engineering picture.
How precipitated silica is separated from solution, how water is evaporated from the NaOH circuit, and how spray drying converts slurry to powder — plus how MVR transforms the energy economics.
After precipitation, you have a slurry: fine SiO₂ particles suspended in Na₂CO₃ solution. The goal is to separate the SiO₂ solids from the Na₂CO₃ liquid (which goes to causticisation), then wash the SiO₂ cake to remove residual sodium and impurities.
A filter press consists of a series of hollow plates (typically 30–100 plates per press) with filter cloths stretched over each face. The slurry is pumped in under pressure (6–15 bar). Liquid passes through the filter cloth, solids build up as a "cake" between the plates. When the press is full (cake resistance too high), pumping stops, wash water is pushed through to rinse the cake, then the plates separate and the cake drops out.
The filter cloth material is critical: polyester or polypropylene woven fabric with a pore size matched to the SiO₂ particle size. Too coarse → silica passes through (yield loss). Too fine → blinding (pores clog, flow rate drops).
Figure 6.1 — Filter press operating cycle: Fill → Filter → Wash → Open/Discharge
Type: Recessed plate, membrane-assisted (optional)
Plates: 60–100 per press, polypropylene
Filter cloth: Polypropylene monofilament, 5–15 µm pore
Operating pressure: 6–12 bar
Cycle time: 60–90 min (fill+filter+wash+open)
Cake moisture: 50–65% (before drying)
Quantity: 2 presses (Ph1A); +1 press added in Phase 1B
CAPEX: ₹110 Lakhs (2 presses, Ph1A)
After filtration, the SiO₂ cake contains residual Na₂CO₃ and NaOH in the pore water. If not washed, these sodium compounds remain in the product and increase ash content (Na₂O spec is critical for HDS and dental grades).
For dental grade: Na₂O <0.5%, which requires 3–5 wash cycles. Each wash dilutes sodium concentration by ~80%. After 3 washes: 0.2³ = 0.8% of original Na remains.
Wash water (dilute Na₂CO₃) is sent to the evaporator to concentrate before causticisation — not discarded.
The NaOH circuit operates with large volumes of dilute solution. After leaching, the sodium silicate solution is diluted by the 10% NaOH starting concentration plus wash water returns. Before causticisation, the Na₂CO₃ solution must be concentrated to ~15–20% to achieve good causticisation efficiency (higher [Na₂CO₃] → faster reaction, better conversion).
Evaporation uses steam to boil off water from the solution, increasing concentration. At atmospheric pressure, water boils at 100°C. The heat required to evaporate 1 kg of water is the latent heat of vaporisation = 2,257 kJ/kg (at 100°C).
In a single-effect evaporator, steam boils liquid in one vessel, the steam vapour is condensed and the condensate discarded. For every 1 kg of water evaporated, you need ~1.1 kg of steam (accounting for heat losses).
A triple-effect evaporator connects three vessels in series. Steam enters the first vessel (highest pressure, ~5 bar, ~155°C). The vapour produced in the first vessel is used as the heating medium for the second vessel (lower pressure, ~80°C). Vapour from the second vessel heats the third. One unit of steam now evaporates ~2.5–2.8 units of water. This reduces steam consumption to 0.35–0.40 kg steam per kg water evaporated — a 3× efficiency improvement over single-effect.
Fluxara uses a triple-effect evaporator with MVR on the first effect — described in Chapter 8.
Water to be evaporated from NaOH/Na₂CO₃ circuit:
Estimated: ~75,000 kg/day (from process mass balance)
Single-effect steam need:
75,000 kg/day × 2,257 kJ/kg ÷ 1,000 = 169,275 MJ/day ≈ 169 GJ/day
(This is the 167.1 GJ/day "no MVR" evaporation load in the energy balance)
Triple-effect (without MVR, ~2.5× efficiency):
169 ÷ 2.5 = 67.6 GJ/day steam needed
With MVR (65% saving on first effect):
MVR saves 0.65 × 169 = 109.85 GJ/day → actual consumption: 58.5 GJ/day ✓
(matches DPR v15 energy balance Section 04)
After filter pressing, the SiO₂ cake is reslurried with water (or the CTAB solution for HDS grade) to form a pumpable slurry. This is fed to a spray dryer via a high-pressure pump and atomised through a nozzle (or rotating disc atomiser) into a large cylindrical chamber.
Atomisation breaks the slurry into millions of fine droplets (diameter: 50–200 µm). Hot air (130–200°C) is blown co-currently or counter-currently through the chamber. Each droplet loses its water content in 0.5–5 seconds as it falls through the hot air. The dry silica particles exit at the bottom cone, collected by a cyclone separator and bag filter. Exhaust air exits via the top with the evaporated moisture.
Final product moisture: <6% (target). BET surface area is largely determined by precipitation conditions — spray drying preserves it. For dental grade, inlet temperature must be controlled to ≤180°C to avoid surface modification of sensitive silica.
Type: Co-current, rotary atomiser or two-fluid nozzle
Capacity: 500 kg/hr evaporation rate per dryer
Quantity: 2 dryers (Ph1A); +1 added in Phase 1B
Shell material: SS316 (food grade compatible)
Inlet air temperature: 130–200°C
Outlet temperature: 70–90°C (product)
Product moisture: <6% w/w
Heat source: Steam from boiler
CAPEX: ₹160 Lakhs (2 dryers, Ph1A)
SS304 (18% Cr, 8% Ni) is adequate for most process equipment. SS316 adds 2–3% Molybdenum, which provides superior resistance to chloride-induced pitting corrosion.
For food-grade and dental-grade PS: regulatory specifications (IS 6579, FSSAI E551) require that contact surfaces be non-reactive and non-contaminating. SS316 is the food industry standard.
For HDS grade (tyre): trace metal contamination (Fe, Cr) degrades rubber properties. SS316's lower Fe leaching rate matters here too.
The cost premium of SS316 over SS304 is ~15–20% — small relative to the product revenue enabled.
For HDS grade, CTAB is dissolved in water and mixed into the silica slurry before spray drying. As each droplet dries, the CTAB molecules migrate to the silica surface (driven by the receding water front) and self-assemble into a monolayer coating. This is the most efficient way to coat: in-situ during drying, at 0.289 MT/day CTAB consumption.
The alternative (post-drying coating in a mixer) gives less uniform coverage and higher CTAB waste. In-slurry coating during spray drying is the standard industrial method for HDS production.
How precipitated silica and nano-PCC quality is measured, what BET surface area means, why CTAB value determines tyre performance, and what the QC lab at Fluxara must be capable of.
BET stands for Brunauer–Emmett–Teller — the three scientists who developed the theory in 1938. BET surface area is the total surface area of a solid material per gram, measured in m²/g.
The measurement works by adsorbing nitrogen gas (N₂) onto the silica surface at −196°C (liquid nitrogen temperature). As pressure increases, more N₂ molecules adsorb onto the surface. By measuring how much N₂ is adsorbed at different pressures, the total surface area can be calculated (each N₂ molecule occupies a known area: 0.162 nm²).
Why does surface area matter? In tyre rubber, silica acts as a reinforcing filler. The silica particles physically interact with the polymer chains — more surface area = more contact area = stronger reinforcement. But more surface area also means more silanol groups (Si-OH) on the surface, which interact with the rubber. CTAB treatment masks some of these silanols while preserving dispersibility.
Standard grade: 140–165 m²/g BET. HDS grade: typically 160–180 m²/g. Dental grade: 100–140 m²/g (lower surface area preferred for abrasion control in toothpaste).
BET measures all surface area — including micropores that polymer chains cannot physically enter. CTAB surface area measures only the surface accessible to CTAB molecules (which are large enough to approximate a polymer chain's reach — about 0.35 nm). So CTAB surface area is always lower than BET, but is a better predictor of in-rubber performance.
HDS specification: CTAB ≥175 mg CTAB per gram silica. This is determined by measuring how much CTAB adsorbs from solution onto a known mass of silica (titrimetric method). Higher CTAB value = more polymer-accessible surface = better green tyre performance.
Particle size distribution is measured by laser diffraction: a laser beam passes through a suspension of particles in a liquid, and the diffraction pattern reveals the size distribution. Results are reported as D10, D50, D90:
D50 (median particle size): 50% of particles are smaller than this value. The central measure. Standard PS: D50 ≤20 µm. Dental PS: D50 ≤12 µm.
D90: 90% of particles are smaller than this value. Indicates the coarse tail of the distribution. D90 ≤40 µm is typical for HDS grade — coarse particles cause tyre defects.
D10: 10% are smaller. Indicates fine particles. Very fine particles can cause dustiness and handling issues.
For dental/toothpaste grade, tighter control is required: D50 ≤12 µm AND no particles >45 µm (sieve test). Coarse particles would scratch tooth enamel.
X-Ray Fluorescence (XRF) bombards the sample with X-rays, causing each element to emit fluorescent X-rays at characteristic energies. By measuring these energies and intensities, you can identify and quantify every element present >0.01%.
At Fluxara, XRF is used for:
1. Incoming RHA QC: Confirm SiO₂ ≥92%, check for high Fe, Ti, or other contaminants from the supplier's process.
2. CaO purity: Confirm ≥85% CaO. Also check MgO (high MgO gives "dead burned" lime that slakes poorly) and SiO₂ (limestone contamination).
3. Finished PS purity: Dental grade requires Pb ≤1 ppm, As ≤1 ppm, heavy metals total ≤20 ppm. XRF at 1 ppm detection level requires energy-dispersive XRF (ED-XRF) — the ₹40L QC lab allocation covers this instrument.
| QC Test | Instrument | What It Measures | Grade Critical | Frequency |
|---|---|---|---|---|
| BET Surface Area | BET Analyser (N₂ adsorption) | Total surface area m²/g | All PS grades | Every batch |
| CTAB Value | UV-Vis spectrophotometer + titration | Polymer-accessible surface area | HDS mandatory | Every HDS batch |
| Particle Size (PSD) | Laser diffraction (e.g. Mastersizer) | D10, D50, D90 | HDS, Dental | Every batch |
| XRF elemental | ED-XRF spectrometer | % SiO₂, Fe, Pb, As, heavy metals | Dental (Pb ≤1 ppm) | Every delivery + product |
| Moisture content | Moisture analyser / Karl Fischer | % H₂O | All grades | Every batch |
| pH | pH meter | Solution pH | All grades | Continuous (in-line) |
| Oil absorption | Spatula method (ASTM D281) | mL oil / 100g silica | Rubber grades | Weekly |
| Dissolution rate | NaOH dissolution test | Amorphous content of RHA | Feedstock QC | Every RHA delivery |
Table 7.1 — QC test matrix for Fluxara products and raw materials
| Grade | d50 Target | Key Test |
|---|---|---|
| Coatings / Bulk | 2 µm | Brightness, d50 PSD |
| Sealant (stearic) | 0.7 µm | d50, oil absorption, stearic % |
| Plastics (OCC) | 0.5 µm | d50, dispersibility index |
Stearic acid coating percentage is tested by TGA (Thermogravimetric Analysis) — heating the coated PCC to 600°C under nitrogen. The organic coating burns off at ~300–400°C, measured as weight loss = coating %.
Target: 1.5–2.5% stearic acid on PCC surface. Too little: particles remain hydrophilic, don't disperse in sealant polymer. Too much: wasted reagent, surface saturation, no further benefit.
The full energy picture — how 83 GJ/day of surplus heat is generated, why MVR is the transformative technology, and how Phase 1B and Phase 2 are anchored to this surplus.
In a conventional evaporator, you supply steam at high temperature → it condenses, releasing heat → this heat evaporates water from the process liquid → the water vapour (at 100°C, atmospheric pressure) is condensed and discarded, taking its latent heat with it. This is thermodynamically wasteful: you paid for high-grade steam energy, used only a fraction of it (the temperature difference), and threw away most of the latent heat.
MVR recovers the discarded latent heat by recompressing the water vapour. A mechanical compressor (driven by an electric motor, ~150–200 kW) takes the vapour produced at 100°C and compresses it to a higher pressure (and therefore higher temperature — typically 115–120°C). This compressed vapour is now hot enough to serve as the heating medium for the same evaporator body — so the evaporator heats itself. Only a small electrical input is needed to run the compressor, rather than large quantities of steam.
Figure 8.1 — Conventional vs MVR evaporator: vapour recompression saves 65% of evaporation steam
Without MVR:
Evaporation steam demand: 167.1 GJ/day
Total process steam demand: 167.1 + 42.7 (leach) + 15.1 (spray dry) = 224.9 GJ/day
Steam available: 199.3 GJ/day
DEFICIT: 224.9 − 199.3 = 25.6 GJ/day ← plant CANNOT run at design throughput
With MVR (65% saving on evaporation):
MVR saving: 0.65 × 167.1 = 108.6 GJ/day
Evaporation with MVR: 167.1 − 108.6 = 58.5 GJ/day
Total process steam: 58.5 + 42.7 + 15.1 = 116.3 GJ/day
Steam available: 199.3 GJ/day
SURPLUS: 199.3 − 116.3 = 83 GJ/day ✓
MVR electric motor power (to run the compressor):
Approximate: 150–200 kW continuous
Annual: 175 kW × 8,760 hr × 0.85 availability = ~1.3M kWh/yr
Cost: 1.3M × ₹7/kWh = ₹0.91 Cr/yr electricity
Steam equivalent saved: 108.6 GJ/day × 330 days = 35,838 GJ/yr
Value of steam saved: 35,838 GJ ÷ 2.4 MJ/kg steam × ₹... (included in base OPEX)
Net MVR benefit: saves ~₹12 Cr/yr in steam + enables plant to operate at all
The 83 GJ/day surplus exists because rice husk has more energy than the process needs — even with the inefficiency of an 82% boiler and the heat demands of leaching + drying. This is by design: the feedstock (rice husk) was specifically chosen because its energy content exceeds the process heat requirement, generating surplus for expansion.
Phase 1B (9 GJ/day) and Phase 2 (~74 GJ/day remaining) are anchored to this surplus. This means no additional fuel purchase is ever needed for plant expansion — the fuel is the rice husk already being combusted. This is a critical capital efficiency advantage: Phase 2 gets its energy for free.
| Heat Allocation | GJ/day | Notes |
|---|---|---|
| Useful heat from boiler | 199.3 | 243 × 82% |
| Leach reactor heating | −42.7 | Heating 15 MT/day RHA slurry to 90°C |
| Evaporation (with MVR) | −58.5 | MVR reduces 167.1 to 58.5 GJ/day |
| Spray drying | −15.1 | Drying PS to <6% moisture |
| Phase 1A surplus | 83.0 | 199.3 − 116.3 = 83 GJ/day |
| Phase 1B (extra 4.568 MT/day RHA) | −9.0 | Equipment-constrained, not thermally |
| After Phase 1B surplus | ~74.0 | Phase 2 thermal anchor |
How equipment is sized, the civil and infrastructure requirements, the 3-acre site layout, and the rationale for each major design decision.
Every piece of equipment is sized from the mass balance (Ch.3) plus a design margin of typically 15–25%. The design margin accounts for: feed variability, peak demand periods, cleaning downtime, and future debottlenecking.
Example — Leach Reactor: Total RHA to leach: 15.032 MT/day. At 2-hour residence time and 10% solids loading: volume of slurry = 15,032 kg ÷ 0.10 = 150,320 kg slurry = ~150 m³ needed per 2 hours. Three reactors of 10,000 L each = 30 m³ total active volume — this is for semi-continuous operation where reactors are staggered so one is always filling, one reacting, one discharging. With a 25% margin, 3 × 10 KL is adequate.
| Equipment | Sizing Basis | Design Capacity | CAPEX ₹L |
|---|---|---|---|
| Furnace (grate-fired) | 20 MT/day husk at ≤700°C | 1.0–1.4 MT/hr continuous | 280 |
| IBR Boiler (fire-tube) | 199.3 GJ/day useful heat | ~3,500 kg/hr steam @ 8 bar | (inc. above) |
| Leach Reactors (×3, 10KL) | 15.032 MT/day RHA, 2-hr RT | 30 m³ active volume | 120 |
| Filter Presses (×2) | 13.149 MT/day PS cake | 60–90 min cycle, 2 in parallel | 110 |
| Spray Dryers (×2, 500 kg/hr) | 13.149 MT/day PS at <6% moisture | 1,000 kg/hr total evaporation | 160 |
| Triple-Effect Evaporator + MVR | 75,000 kg/day water evaporation | MVR-assisted, 58.5 GJ/day | 150 |
| Causticisation + CaO Slaker | 13.364 MT/day CaO, 20.272 MT/day PCC | Agitated vessels, steam-jacketed | 90 |
| PCC Hydrocyclone Classifiers | 20.272 MT/day PCC at d50 0.5–2 µm | Multi-stage classification | 80 |
| ETP + ZLD (25 KLD) | 84 KLD fresh water + 80% recycle | 25 KLD zero liquid discharge | 120 |
| QC Lab (BET + PSD + XRF) | All grade testing as per Ch.7 | Full analytical suite | 40 |
Figure 9.1 — Schematic site layout for 3-acre TSIIC Sangareddy site (not to scale)
The furnace and boiler structure must be fire-rated RCC (Reinforced Cement Concrete) — not Pre-Engineered Building (PEB). Reasons: (1) IBR regulations require masonry or RCC for boiler houses. (2) Rice husk storage is a fire hazard — the boiler building must provide fire compartmentation. (3) The CEMS chimney requires a reinforced concrete foundation capable of withstanding wind loads on a 15–20m stack.
RCC cost premium over PEB: ~30–40%. Justified by regulatory compliance, fire safety, and the 20+ year structural life required for a financed facility.
TSPCB mandates ZLD for chemical manufacturing plants at Sangareddy (MIDC/industrial zone). ZLD means no wastewater may be discharged to drains, waterways, or land. All effluent must be treated and recycled or evaporated to a solid residue.
Fluxara's 25 KLD ETP+ZLD system includes: equalization tank → chemical precipitation (heavy metals) → MBR (membrane bioreactor for organics) → RO (reverse osmosis, 80% recovery) → MEE (multiple effect evaporator for RO reject) → solid residue for disposal. The 80% ZLD water recovery figure drives the 84 KLD fresh water makeup calculation.
How the plant is controlled, monitored, and protected — from pH control in precipitation to SCADA, interlock systems, and the safety philosophy for a chemical plant with NaOH, CaO, and CO₂.
Precipitated silica quality (grade, surface area, particle size) is determined by tightly controlled process conditions: pH during precipitation, temperature profiles, residence times, and reagent ratios. A 0.5 pH unit deviation during CO₂ addition can shift product from HDS grade (₹45/kg) to standard grade (₹26/kg) — a ₹19/kg revenue loss on every kilogram of that batch.
The plant operates 24/7 in 3 shifts. Manual control is insufficient for: (a) the precision required (pH ±0.2 units), (b) the consistency across shifts and operators, (c) safety interlocks that must act faster than a human can respond. A SCADA (Supervisory Control and Data Acquisition) system is essential from Day 1.
| Process Variable | Control Method | Setpoint | Consequence of Deviation |
|---|---|---|---|
| Furnace temperature | CEMS + auto-shutoff | ≤700°C (shutoff at 720°C) | Over 700°C: cristobalite → batch loss |
| Leach reactor temperature | Steam control valve (PID) | 90°C ±2°C | Below 80°C: poor extraction; above 95°C: NaOH volatilisation |
| Leach reactor pH | In-line pH probe + NaOH dosing valve | pH 12–13 | Below 11: incomplete extraction; above 14: scale formation |
| Precipitation pH | CO₂ flow control (mass flow meter) | pH 5.5–9 (grade-dependent) | Wrong pH → wrong grade → revenue loss |
| Precipitation temperature | Cooling water control valve | 50–70°C | Above 80°C: silica redissolves; below 40°C: slow kinetics |
| Boiler steam pressure | Pressure controller (PIC) | 6–8 bar(g) | Low pressure: process temperatures drop |
| CaO slaker temperature | Cooling water jacket valve | 80–90°C | Above 100°C: safety risk (steam flashing) |
| Filter press cycle | PLC timer + pressure sensor | Fill: 20 min, Filter: 30 min, Wash: 20 min | Over-filling: cloth damage; under-washing: Na contamination |
In-line glass electrode pH sensors. Critical in leach reactors and precipitation vessels.
Calibrate daily with buffer solutions (pH 4, 7, 10). Replace electrode every 3–6 months (glass degrades in NaOH).
Failure mode: drifting reading → wrong pH → grade deviation. Always have spare electrode.
Pt100 RTDs for process vessels (±0.5°C accuracy). Type-K thermocouples for furnace (up to 1200°C range). CEMS temperature sensor for chimney gas (triggers auto-shutoff).
RTDs require 4-wire connection for accuracy. Check calibration quarterly against certified reference.
Mass flow meters (Coriolis type) for CO₂ gas feed — critical for precipitation stoichiometry.
Magnetic flow meters for slurry (NaOH, slurry) — no moving parts, suited for abrasive/corrosive fluids.
Rotameters for cooling water, low-precision flows.
NaOH (Sodium Hydroxide) — Corrosive, GHS05: Causes severe chemical burns to skin, eyes, and respiratory tract. At 10% concentration (process), contact causes burning within seconds. At 48% (concentrate deliveries), contact is immediately destructive. All NaOH handling areas require: safety showers + eyewash stations within 10 seconds walking distance, chemical-resistant PPE (nitrile gloves, face shield, apron), ventilation. NaOH is not flammable but reacts with aluminium to produce hydrogen gas (avoid aluminium fittings in NaOH lines).
CaO (Quicklime) — Corrosive, Reactive with Water: Slaking is exothermic (63.7 kJ/mol). Contact with moisture (including skin perspiration) causes localised heating and alkaline burns. CaO delivery and slaker operation require: dust respirator, face shield, heat-resistant gloves. The slaker vessel must be vented to remove steam/aerosols. CaO is not flammable.
CO₂ (Carbon Dioxide) — Asphyxiant: CO₂ from the furnace is fed to precipitation vessels. At concentrations above 5%, CO₂ displaces oxygen and causes unconsciousness (it is colourless and odourless). Precipitation areas must have continuous CO₂ gas detectors with alarms at 1% (warning) and 3% (evacuate). CO₂ is heavier than air — detectors at floor level.
Rice Husk — Fire and Dust Hazard: Rice husk dust in air is combustible (explosive range ~30–120 g/m³). Husk storage areas require fire suppression (water sprinklers), no ignition sources (no sparks, no smoking), and dust control (enclosed conveyors). Emergency water supply to husk storage must be available within 30 seconds.
| Hazard | Location | Primary Control | Emergency Response |
|---|---|---|---|
| NaOH spill/splash | Leach reactors, NaOH storage | Bunded area, PPE, safety shower | Flush with water 15+ min, medical aid |
| CaO dust inhalation | Slaker, CaO unloading | Enclosed handling, P3 respirator | Move to fresh air, wash eyes, medical |
| CO₂ asphyxiation | Precipitation area, basement zones | Continuous CO₂ monitor, ventilation | Evacuate, fresh air, rescue breathing |
| Steam burn (boiler) | Boiler house, steam lines | Insulation, pressure relief valves, IBR cert | Cool with water, do not remove clothing, medical |
| Husk fire | Husk storage | Sprinklers, no ignition, firebreak | Water suppression, TSPCB notification |
| Slaking exotherm | CaO slaker | Cooling jacket, temperature alarm | Shut CaO feed, increase cooling |
SCADA (Supervisory Control and Data Acquisition) collects sensor data from the entire plant, displays it on operator screens, executes control loops (PID), and logs data for quality records and regulatory compliance.
For Fluxara's vision of remote oversight: A SCADA system with secure remote access (VPN + encrypted connection) allows the plant manager or owner to monitor all process parameters from anywhere — mobile phone or laptop. The foundation for the Digital Twin and AI/ML layer described in the long-term vision is SCADA data collection. Every sensor reading (pH, temperature, flow, pressure) is timestamped and stored — this historical data is the training set for future ML models.
PLC (Programmable Logic Controller): Allen-Bradley or Siemens S7 series
HMI: 21" touchscreen in control room + remote web interface
Tags: ~200 I/O points (temperature, pH, flow, pressure, level)
Historian: 5-year data retention (TSPCB compliance)
Alarm management: 3-tier (warning, alarm, emergency)
Integration: CEMS data feed to TSPCB portal (statutory)
SCADA data → time-series database (InfluxDB/TimescaleDB)
Process model: digital replica of mass/energy balance → deviation alerts
ML models: yield prediction from pH + temp + RHA quality inputs
Anomaly detection: flag when sensor readings deviate from expected pattern
Predictive maintenance: motor current signature → bearing fault prediction
Remote dashboard: Grafana or custom web app → overseer can view KPIs from Telegram
The path from Day 1 SCADA to full remote AI oversight: (1) Wire all sensors to PLC from Day 1 — this costs little extra but creates the data infrastructure. (2) Stream all data to a time-series database (InfluxDB runs on a ₹15,000 mini-server). (3) Build a Grafana dashboard with KPI panels — accessible from phone. (4) Add MAIS agent that queries the database and sends daily reports to Telegram. (5) Train ML models on historical data (after 6 months of operation) to predict silica grade from process variables. (6) Digital Twin: Python model that runs a live simulation in parallel with the real plant, flagging deviations.
Steps 1–3 cost under ₹5L extra at commissioning. Steps 4–6 are software-only, built progressively. This is the architecture that enables you to oversee the plant from anywhere in the world via a Telegram chat.
From the molecular structure of rice husk silica to the SCADA architecture that will eventually run the plant remotely — 10 chapters of ground-up engineering theory, worked examples, and plant-specific numbers.
| Chapter | Topic | Key Number to Remember |
|---|---|---|
| 01 | Raw Materials Science | RHA: ≥92% SiO₂ · amorphous · dissolves in NaOH at 90°C · ≤700°C strict |
| 02 | Combustion & Furnace | 20 MT husk × NCV 12,150 kJ/kg = 243 GJ gross → 199.3 GJ useful @82% |
| 03 | NaOH Leaching | 1.331 kg NaOH per kg SiO₂ · 90°C · 2hr · PP-lined SS · 88% extraction |
| 04 | Silica Precipitation | pH 5.5–7 = dental · pH 6–8 = HDS · pH 8–9 = standard · CO₂ is free |
| 05 | Causticisation & PCC | Na₂CO₃ + Ca(OH)₂ → 2NaOH + CaCO₃↓ · 82% recovery · PCC is obligatory |
| 06 | Separation & Drying | Filter press: 60–90 min cycle · Spray dryer: <5 sec drying · SS316 mandatory |
| 07 | Product Quality | BET m²/g · CTAB ≥175 mg/g (HDS) · D50 ≤12 µm (dental) · Pb ≤1 ppm |
| 08 | Energy & MVR | MVR saves 108.6 GJ/day → 83 GJ/day surplus · without MVR plant cannot run |
| 09 | Equipment & Layout | 3 acres · fire-rated RCC boiler house · ZLD Day 1 · QC lab ₹40L |
| 10 | Control & Safety | SCADA from Day 1 · pH ±0.2 units critical · CO₂ monitors mandatory · data = future ML |
With this engineering foundation, the next phase of learning covers: (1) Building the Digital Twin — a Python simulation model that mirrors the plant's mass and energy balance in real time. (2) IoT sensor architecture — which sensors, what protocols (Modbus, OPC-UA), how to get data from the plant floor to the cloud. (3) ML models for yield optimisation — using pH, temperature, RHA quality, and flow rates to predict silica grade and extraction efficiency. (4) MAIS integration — connecting the Digital Twin to the Telegram bot so you can ask "what is the current leach reactor temperature?" and get a live answer.
These are the next documents in the series. The engineering course (Chapters 1–10) is the prerequisite — you cannot build useful ML models without understanding what you are modelling.