Purified phosphoric acid: TBP extraction, scrub and strip — a BRINE process flowsheet
Food-, technical- and (downstream) battery-grade acid is not made by cleaning wet-process acid of its metals; it is made by extracting the ACID out and leaving the metals behind. Three countercurrent stages of 80 vol% tri-n-butyl phosphate in kerosene take the H₃PO₄ out of a sulfate-salted 35 wt% feed, three scrub stages against product acid wash the co-extracted impurities back out of the loaded organic, water strips the acid into a clean aqueous liquor, and an evaporator concentrates it. The acid's own distribution ratio is NOT typed in: it comes from Alibrahim's measured single-contact yields (Periodica Polytechnica Chemical Engineering 51(1) 2007 39-42, doi:10.3311/pp.ch.2007-1.07, 80 vol% TBP, equal phase volumes, room temperature) at the feed's own strength, converted from that mass basis to the cascade's mole basis with both phases' mean molar masses — a factor of about ten for this system, so not optional. Sulfuric acid is the salting agent because the paper measured it as the strongest (H₂SO₄ > HCl > HNO₃ > HClO₄) and because a phosphoric plant already has it: 0.8 M lifts the single-contact yield from 12.3% to 16.9%. On that basis alone, three stages at a solvent-to-feed ratio of seven recover 83.9% of the acid, which is the paper's own McCabe-Thiele conclusion ('2 to 3 stages extract 80%') reached from its equilibrium data rather than fitted to its answer. Solved: 10 kg/s of feed acid gives 3.39 kg/s of product at 85.8 wt% H₃PO₄ (62.2 wt% P₂O₅) carrying 7 ppm Fe, 11 ppm Ca, 0.5 ppm Zn, 260 ppm S and 1.0% F, with 83.3% of the fed acid reaching it and the rest leaving in an 8 wt% raffinate that goes back to fertilizer-grade acid. Those impurity figures are the paper's own product analysis, reproduced — which is the point and also the caveat below.
Modeling assumptions & limitations
- 1And they matter:
- 2NO metal distribution ratio is published with that data, so none ships — the metals' K_D in the scrub and strip here are BACK-FITTED to the paper's own measured feed-to-product decontamination over its own 3+3-stage circuit (Fe 1850 → 7 ppm, Ca 613 → 11, Zn 204 → nil, S 7350 → 260), so the impurity rejection is a reproduction of that measurement, not a prediction, and a different acid needs its own leaf test
- 3The model carries no extractant loading, so recovery is overstated at a high organic loading
- 4Fluorine is barely rejected here, which is the real behaviour — the paper's product still held 1% F against 2% in the feed, and a real plant strips it with steam during concentration
- 5The brine package prices a hydrocarbon phase by falling back on its dominant component, so the organic streams' enthalpies are indicative and the cascade is a material balance
- 6The solvent's water uptake, which sets how much water the strip returns, is not measured here
- 7The concentrator is one effect and its product strength is steeply sensitive to its duty — 10.0 kg/s of steam gives 81 wt% and 10.5 gives 88 — because the last water is the hardest to remove; a real plant concentrates in several effects under tighter control, which this repo's phosphoric-acid-concentration-* examples show, and above 95 wt% the acid is superphosphoric and outside the water-activity model's range.
26 views 0 forks
- Rigorous BRINE thermodynamics, solved by the same engine every simulation runs on.
- 2 unit operations modeled: 3× Strip, Concentrator.
- Focus areas: Purified phosphoric acid, Solvent extraction, TBP, Food grade.
- Thermodynamics
- BRINE
- Components
- water, phosphoric_acid, h2so4, hydrogen_fluoride, fe3, ca, zn, tri_n_butyl_phosphate, n_dodecane
- Unit operations
- 3× StripConcentrator
Opens in a new tab, loaded straight into the app — no setup.
Read the step-by-step guideReproduce this exact result from Python — the real client.get_example() → run_and_wait() path, not a mockup.
from flowsim.sdk import FlowSimClient
client = FlowSimClient()
example = client.get_example("purified-phosphoric-acid-tbp")
sim = client.create_simulation(example["title"], example["flowsheet"])
result = client.run_and_wait(sim["id"])
print(result["status"]) # "converged"
streams = client.streams(sim["id"])Related models
Cobalt/nickel solvent extraction (D2EHPA)
A laterite leach liquor (Co/Ni/Mg in dilute sulfate solution) meets a D2EHPA-in-kerosene organic phase across a 20-stage countercurrent extraction circuit run at pH 5.2 — the real industrial operating window (e.g. Bulong, Murrin Murrin, Western Australia) that exploits the ~1 pH-unit gap between Co and Ni's D2EHPA extraction isotherms: Co extracts into the organic while Ni (and Mg gangue) are rejected to the raffinate. The loaded organic then meets fresh dilute-acid strip liquor across a 10-stage strip circuit at pH 1.0 (well below Co's isotherm), reversing the equilibrium to recover a concentrated cobalt strip liquor and regenerate barren organic. Real pH-isotherm chemistry throughout, not a fitted shortcut K_D.
Battery black-mass recycling: leach -> SX -> BPED -> LiOH crystallization
Spent-battery black mass (an NMC111-like LiNi1/3Mn1/3Co1/3O2 lump, sulfuric acid-leached at a screening-level yield/stoichiometry — see the stated scope) dissolves into a pregnant leach solution carrying Li/Co/Ni/Mn. A 15-stage D2EHPA solvent-extraction circuit (the same real pH-isotherm chemistry the co-ni-solvent-extraction example uses, operated at pH 7.5 — above every curated Co/Ni/Mn pH50 — so all three transition metals extract into the kerosene organic phase while Li, which has no curated D2EHPA isotherm, stays in the aqueous raffinate untouched) purifies the liquor before it ever reaches the lithium-recovery chemistry. A Faraday's-law bipolar-membrane electrodialysis (BPED) cell (the same real electrochemistry as the direct-lithium-extraction example, sized up for this liquor's larger Li flow) converts Li+ to LiOH, which an MSMPR crystallizer takes to battery-grade LiOH solid. Every step reuses an already-real, independently-tested unit op (,,,) chained into the one train none of them had been assembled into before.
Battery cobalt recovery: leach, neutralize, D2EHPA Mn SX, Cyanex 272 Co SX, electrowinning
Spent NMC111 black mass is leached in sulfuric acid with the acid consumption tracked (lithium as Li₂SO₄, each transition metal as its sulfate), so the pregnant leach solution carries the residual free acid a real leach leaves. A neutralizer then titrates that liquor with 20 wt% caustic to pH 4.0 on the aqueous speciation: the caustic dose is solved, not assumed, and the heat of neutralization (55.8 kJ per mole of acid neutralized) warms the liquor; the caustic header is the supply, and what the titration does not take is returned. Two solvent-extraction circuits on the pH-isotherm chemistry follow, the common NMC arrangement: D2EHPA at pH 4.0 takes manganese into the organic while cobalt and nickel stay behind, and Cyanex 272 at pH 5.0 then takes cobalt and rejects nickel and lithium to the raffinate. The Cyanex 272 isotherms are one published measurement (Liu et al., Metals 11 (2021) 401: 0.4 M in ShellSol D70, 40 C, sulfate; pH50 cobalt 4.15, nickel 6.06), and the 1.9 pH-unit gap between them is why the cobalt organic carries about 0.6% of the nickel, against about 3% with D2EHPA at the same cobalt recovery. The loaded cobalt organic is stripped into a dilute sulfuric strip liquor, and that advance electrolyte goes to cobalt electrowinning: Faraday's law at a 90% current efficiency, 90% of the cobalt plated as cathode cobalt, the anode's oxygen and the 10% of the charge that evolves hydrogen reported as cell gas, and one mole of sulfuric acid regenerated per mole of cobalt plated.
Guemassa-type cobalt refinery (representative): arsenide leach, scorodite, Cu EW, Cyanex 272, Co EW, CoSO4, Ni(OH)2
A representative Bou-Azzer/Guemassa-type cobalt refinery, not Managem's plant. Managem's Bou-Azzer mine supplies most of the cobalt concentrate its Guemassa hydrometallurgical plants turn into cobalt cathodes, cobalt derivatives and nickel (managemgroup.com, read 2026-10-23), and La Vie Eco (30 December 2025, read) reports a cobalt sulphate unit there of up to 5,800 t/y, from Bou-Azzer metal, entering service in 2026. The plant's own flowsheet and operating figures are not public, so every plant number here is labelled representative. The feed is read from a source: the Bou Azzer concentrate Managem supplied for study, 91.9 g/kg Co, 598 As, 84.3 Fe, 17.5 Ni, 1.9 Cu and 8.1 S (Johnson et al., Hydrometallurgy 2020, Table 1), a skutterudite/loellingite arsenide. Its Fe:As molar ratio comes out 0.19, as that paper states, against about 1 needed to fix arsenic as scorodite. 0.80 kg/s of it (about 2,300 t/y of contained cobalt, the district's output in the same paper) is leached in sulfuric acid at 90 C (95% extraction and the even As(III)/As(V) split are representative inputs), giving a liquor of about 25 g/L As. The arsenic_precipitation unit then oxidises the As(III) and Fe(II) with 50% H₂O₂ (3.6 mol/s, the stoichiometric demand, drawn from the supply with the rest returned), tops the iron up with ferric sulfate to Fe:As 1.1 (5.5 mol/s Fe, more iron than the concentrate itself carries, which is why arsenide feeds are usually roasted to arsenic trioxide first), and doses lime to pH 1.8 at 90 C: 1.40 kg/s of scorodite and 1.92 kg/s of gypsum, 0.02 mg/L As left, on a scorodite solubility product of -26.67 at 90 C (Langmuir, Mahoney & Rowson's -25.83 at 25 C carried by the temperature dependence of Zhu, Nordstrom et al. 2017). The residue's equilibrium arsenic in TCLP fluid #1 (pH 4.93, 20 L/kg) is 0.03 mg/L against the 5 mg/L toxicity characteristic. A second stage takes the excess iron out as ferrihydrite at pH 4 with lime. Copper is plated potentiostatically (-0.05 V, plug flow, 92% removed), the liquor is neutralised to pH 5 with caustic, and Cyanex 272 at pH 5 (Liu et al. 2021 isotherms) takes all the cobalt and 0.4% of the nickel. The strip liquor, made up with fresh acid and 75% of the spent electrolyte (recycled), is split: 66% goes to cobalt sulphate, where water is removed (87.5%, a specified removal) and CoSO₄.7H2O crystallises at 30 C to Seidell's solubility (42.26 g CoSO₄ per 100 g water), the mother liquor recycled with a 5% bleed: 0.185 kg/s, about 5,826 t/y, at 95.2% purity with 5% mother liquor in the cake. The rest goes to cobalt electrowinning at 90% current efficiency and 3.2 V, which is 3.23 kWh/kg: the US Bureau of Mines measured 1.48 kWh/lb (3.26 kWh/kg) for cobalt from 75 g/L at 65 C, and about 90% current efficiency. 0.022 kg/s of cathode cobalt (about 692 t/y). Nickel is precipitated from the raffinate as Ni(OH)2 at pH 8.5. Copper is won as cathode, not as copper sulphate: at 1.9 g/kg in the concentrate it never approaches CuSO₄.5H2O saturation in this liquor. Bounds: the amorphous ferric arsenate constant is a 25 C value; no arsenate adsorption on ferrihydrite; no jarosite; the evaporation is a specified water removal (the brine property package carries dissolved cobalt as an inert solute, so a steam-driven effect would boil this liquor to dryness at one temperature) and its duty is not computed; the CoSO₄ solubility is in water, not in the acid mother liquor; the leach, the H₂O₂ utilisation, the copper cell, the splits and the SX flows are representative; the barren organic is not recycled.
Guemassa-type cobalt refinery, roast-leach route (representative): roaster, white arsenic, calcine leach, scorodite polish, Cu EW, Cyanex 272, Co EW, CoSO4, Ni(OH)2
The guemassa-cobalt-refinery flowsheet with a de-arsenication roast in front of the leach, which is how arsenide concentrates are usually treated: the direct leach has to buy more iron than the concentrate carries to fix its arsenic as scorodite. Everything from the scorodite stage on is the same nodes as the direct-leach example. The concentrate is REPRESENTATIVE: the roasting example's mineralogy (12% skutterudite, 8% cobaltite, 30% arsenopyrite, 5% pyrite, 45% quartz) with nickel and copper as NiO and CuO at the Bou Azzer concentrate's Ni/Co and Cu/Co (Johnson et al. 2020), 1.40 kg/s dry with 6% moisture, sized to carry the direct-leach example's 73.5 g/s of cobalt. The fluidized-bed roaster at 700 C and 30% excess air (95% of each arsenide roasted, pyrite 98%, inputs) releases 3.90 MW and rejects 1.46 MW to its coils; the off-gas carries 6.32% SO₂. Cooled to 150 C, 0.46 kg/s of white arsenic condenses (98.84% of the arsenic in the gas, on Welch & Duschak's measured As₄O₆ vapour pressure); in all, 93.9% of the arsenic fed leaves as white arsenic, 1.1% stays in the gas, which goes to a product labelled 'to acid plant' (the acid plant is the roasting example's), and 5.0% stays in the calcine and dust, at 1.84% As. Calcine and dust are leached in sulfuric acid at 90 C, mineral by mineral (inputs: 95% of the Co₃O₄, NiO, CuO and unroasted sulfides and arsenides, 10% of the hematite; Co₃O₄ dissolving as Co(II) assumes a reductive leach whose reductant is not tracked), consuming 2.00 mol/s of acid. The liquor carries 978 mg/L As against 25,228 mg/L in the direct leach, and the iron the calcine dissolves puts its Fe:As at 1.82, so the scorodite stage buys no ferric iron (0.0 mol/s against 5.5 mol/s in the direct leach) and needs 0.18 mol/s of H₂O₂ (3.6), 3.60 mol/s of lime (11.18) and makes 0.05 kg/s of scorodite (1.40) and 0.61 kg/s of gypsum (1.92), leaving 0.03 mg/L As; its TCLP screen is 0.03 mg/L. The iron the scorodite does not need drops as ferrihydrite at pH 4 (0.021 kg/s). Downstream as in the direct-leach example: 95% of the copper plated, 0.5% of the nickel co-extracted by Cyanex 272, cobalt electrowon at 3.23 kWh/kg (0.022 kg/s of cathode), and CoSO₄.7H2O at 0.185 kg/s, about 5,826 t/y at 95.2% purity. Bounds: the roast conversions, dust carry-over, collection efficiency and leach fractions are inputs; skutterudite and cobaltite run on DFT heats of formation (see the roasting example); nickel and copper enter as oxides because the roaster has no Ni or Cu minerals, so the arsenic of the nickel arsenides a real Bou Azzer concentrate carries is missing; oxygen in the leach is not tracked; the SO₂ is not converted here.
Ethylene glycol plant: closed water loop + multi-effect evaporator dehydration
The water-integrated evolution of the fiber-grade MEG plant. Two changes turn the once-through EO/glycol chain into a real, water-economical process: (1) the recovered process water is RECYCLED back to the hydration reactors through a purge splitter. A pure recycle is inventory-singular (reactors are keyed on the shrinking EO pool, so water consumption is fixed regardless of how much water circulates), so an 8% purge pins the loop and makes it well-posed, cutting fresh water makeup from 10 to ~1.75 mol/s (a >80% reduction). (2) The bulk dehydration is done by a genuine TWO-EFFECT EVAPORATOR TRAIN (forward-feed, real steam economy: effect 1's low-temperature vapor is the heating steam for effect 2) doing rigorous (P,H)-flash water removal, rather than a single spec-based split. The recycle converges through the solver's Wegstein tearing (~15 outer passes) and MEG still comes out fiber-grade (>=99.9%). HONEST SCOPE: the reaction chemistry is rigorous stoichiometry (real atom balances, ~90/9/1 selectivity) and the evaporators are real energy-balanced flash effects — but flash evaporation CANNOT reach glycol dryness without slipping glycol into the overhead vapor (MEG has a real vapor pressure at 90 C), so the evaporators run cool and only pre-concentrate. An overhead knockout returns the ~1-2% slipped glycol to the product (no yield loss), and a final vacuum-refining polish (still a spec-based split, representing the refluxed dehydration column a flash cannot replicate) removes the last water. This mirrors a real MEG plant's multi-effect-evaporator + vacuum-refining dehydration section. The product columns remain spec-based (a converged 99.9% fiber-grade MEG column is not tractable under Peng-Robinson here — the MEG/DEG relative volatility is too narrow for the wide-boiling MESH path).