Model
Model
Model
Pick a model. Drill in.
PROMMIS-Core service
IDAES + Pyomo process flowsheets from NETL's PROMMIS library, which spans flotation, leaching, solvent extraction, ion exchange, membranes and brine, roasting, precipitation, comminution, magnet recycling and costing. 18 flowsheets are registered on this service. Live IPOPT solve is the default tier (~seconds); bake / cache / surrogate land in later phases. Refusals are 200 OK responses with a code, not error paths.
Flowsheet coverage
cmi
headlinend_recovery_pct, nd2o3_product_mol_per_s
CMI (Critical Materials Innovation) rare-earth-magnet recycling: Nd2Fe14B feed → copper-nitrate dissolution → S/L filter → pH adjustment → oxalate precipitation → S/L filter → calcination to Nd2O3. A hydrometallurgical alternative to UKy's primary recovery and to hydrogen decrepitation's physical pre-treatment.
paramsfeed_nd2fe14b_mol_per_s, feed_cu_2plus_mol_per_s, feed_h2o_mol_per_s, feed_temperature_k, nd_magnet_conversion, calcination_gas_temp_k
kpisrepm_feed_mol_per_s, nd2o3_product_mol_per_s, nd_recovery_pct
metalsNd, Fe
diafiltration
headlineli_permeate_recovery_pct, co_permeate_recovery_pct, li_co_selectivity
Two-salt diafiltration (LiCl + CoCl2) through a single membrane module. Feed + diafiltrate mix, contact the membrane, and split into a retentate (Co-enriched) + permeate (Li-enriched). The separation target is high Li recovery in the permeate with low Co contamination.
feed_flow_volume, diafiltrate_flow_volume, feed_conc_li, feed_conc_co, feed_conc_cl, diafiltrate_conc_li … (+2)
Get started
These run live IDAES + IPOPT solves through the agent. The first few are quick single-point solves; the later ones sweep, optimize, or compare recovery routes.
Workbench
POST /api/prommis/askParameters
Leave any field blank to use the upstream default.
KPIs
Pick a subset, or leave all unchecked to return every KPI.
kpisretentate_flow_volume, permeate_flow_volume, retentate_conc_li, retentate_conc_co, retentate_conc_cl, permeate_conc_li … (+5)
metalsLi, Co
hydrogen_decrepitation
headlineproduct_mass_kg_per_hr, nd_mass_frac_product_pct, total_heat_duty_kw
Hydrogen decrepitation of end-of-life NdFeB permanent magnets from HDDs: 2.5"/3.5" drives → shredder → H2-atmosphere furnace at ~170°C / 1 atm. The H2 ingress fractures the magnet matrix into friable powder, the canonical first step in REPM recycling. Reports Nd2Fe14B recovery, H2 consumption, furnace size, and heat duty.
paramsflow_2_5inch_hdds_per_hr, flow_3_5inch_hdds_per_hr, operating_temperature_k, gas_inlet_pressure_pa, decrepitation_duration_s, plant_basis_year
kpisrepm_feed_kg_per_hr, product_mass_kg_per_hr, nd_mass_frac_product_pct, h2_consumption_mol_per_hr, furnace_chamber_volume_m3, total_heat_duty_kw
metalsNd, Fe, B
li_co_diafiltration_cost_uncertainty
headlineseparation_cost_mean_usd_per_kg, separation_cost_p10_usd_per_kg, separation_cost_p90_usd_per_kg
Cost-risk study of the three-stage Li/Co membrane diafiltration plant for battery-recycling leachate (100 m3/h, 1.7 kg/m3 Li, 17 kg/m3 Co). The minimum-cost membrane design is found at nominal economics, then re-optimized for each sampled scenario of ten assumptions: membrane price and replacement rate, electricity price, pump efficiency, capacity factor, Lang factor, Li and Co prices (lognormal fits to 2021 price histories), income tax and royalty. Reports the mean, 10th and 90th percentile separation cost in USD_2021 per kg of Li plus Co, the all-in cost with charges on sales (tax, royalty, sales overhead), and each input's correlation with cost. Corrects several PROMMIS defects (Li/Co prices a million times too small, five inputs that never reached the model, inflated tax rates, order-dependent samples).
paramsmembrane_sieving_co, membrane_sieving_li, uq_sample_count, uq_random_seed, uq_sampling_method
kpisseparation_cost_mean_usd_per_kg, separation_cost_p10_usd_per_kg, separation_cost_p90_usd_per_kg, separation_cost_mean_standard_error_usd_per_kg, separation_cost_deterministic_usd_per_kg, all_in_cost_mean_usd_per_kg … (+10)
metalsLi, Co
li_co_diafiltration_costed
headlinecost_of_recovery_usd_per_kg, li_recovery_pct, co_recovery_pct
Battery-recycling leachate (Li and Co in solution) enters stage 3 of a three-stage membrane diafiltration cascade with a wash-water diafiltrate. Li passes the membrane preferentially and leaves as the stage 3 permeate (the Li product); Co is held back and leaves as the stage 1 retentate (the Co product). QGESS plant costing prices the membranes, their replacement, the pressure-drop electricity and the feed and diafiltrate pumps (Lang factor 2, 8064 operating h/yr) and reports the cost of recovery in USD_2021 per kg of Li plus Co. Solve evaluates a given membrane design; optimize resizes the three stages to minimize that cost.
paramsleachate_co_kg_per_m3, leachate_li_kg_per_m3, leachate_feed_m3_per_hr, diafiltrate_wash_m3_per_hr, membrane_sieving_li, membrane_sieving_co … (+6)
kpiscost_of_recovery_usd_per_kg, li_recovery_pct, co_recovery_pct, li_recovery_shortfall_pct, co_recovery_shortfall_pct, li_product_purity_pct … (+12)
metalsLi, Co
li_mg_brine_nanofiltration
headlineli_recovery_pct, mg_passage_pct, permeate_mg_li_ratio_kg_per_kg
Single-stage nanofiltration of lithium brine: Salar de Atacama brine (1.19 g/L Li, 7.31 g/L Mg as chlorides) -> feed pump -> WaterTAP nanofiltration membrane (Donnan steric pore model with dielectric exclusion) -> lithium-bearing permeate and a retentate. The headline KPI is the share of the brine's lithium that reaches the permeate; the magnesium passage and permeate Mg:Li mass ratio show how much separation one stage achieves (little: Mg passes almost as freely as Li here). Pump power and specific energy give the energy cost; there is no capital costing. The square solve is a mid-recovery design (500 m2 at 8 bar); optimize finds upstream's best design at the recovery cap.
paramsnf_membrane_area_m2, nf_pump_pressure_bar, brine_li_g_per_l, brine_mg_g_per_l, brine_water_kg_per_s, nf_max_recovery_frac … (+3)
kpisli_recovery_pct, mg_passage_pct, permeate_mg_li_ratio_kg_per_kg, feed_mg_li_ratio_kg_per_kg, feed_li_g_per_l, feed_mg_g_per_l … (+11)
metalsLi, Mg
membrane_cascade
headlinetotal_annualized_cost, li_precipitate_recovery_pct, co_precipitate_recovery_pct
Multi-stage Li/Co diafiltration cascade with precipitators. Unlike the single-module 'diafiltration' flowsheet, this is a cost-minimization: the cascade DOF is unfixed and IPOPT picks the cheapest operating point (membrane areas, feed splits) that clears the requested Li/Co precipitator-recovery floors. Tune cascade size (stages/tubes), feed, and recovery targets.
paramsnum_stages, num_tubes, flux, feed_solvent_m3_per_hr, feed_li_kg_per_hr, feed_co_kg_per_hr … (+6)
kpistotal_annualized_cost, li_precipitate_recovery_pct, co_precipitate_recovery_pct, li_membrane_recovery_pct, co_membrane_recovery_pct, num_stages
metalsLi, Co
mountain_pass_flotation
headlineoverall_reo_recovery_pct, concentrate_reo_grade_wt_pct
Froth flotation of Mountain Pass bastnaesite ore at plant scale (1700 t/d of dry ore at 7.7 wt% rare-earth oxide). Conditioned pulp -> 12-cell rougher -> cleaner 1 -> cleaner 2 -> cleaners 3 and 4 -> rare-earth concentrate. Cleaner tails recycle upstream, and cleaner-1 tails pass a scavenger whose concentrate is reground and returned to the rougher. Each bank's recovery of rare-earth oxide, calcite, barite, celestine and inert gangue follows first-order flotation kinetics, with residence time set by cell volume, air holdup, pulp density and throughput, so the circuit responds to tonnage, equipment size and reagent effectiveness. The headline KPIs are overall REO recovery to the final concentrate and the concentrate REO grade. Rare earths are one lumped oxide, not individual elements.
paramsflotation_feed_ore_t_per_day, flotation_feed_reo_grade_wt_pct, rougher_reo_rate_multiplier, rougher_cell_volume_m3, rougher_pulp_solids_mass_frac, rougher_air_holdup_frac … (+2)
kpisoverall_reo_recovery_pct, concentrate_reo_grade_wt_pct, concentrate_mass_pull_wt_pct, concentrate_reo_kg_per_hr, concentrate_dry_solids_kg_per_hr, tailings_reo_grade_wt_pct … (+6)
metalsCe, La, Nd, Pr
mountain_pass_flotation_fixed_recovery
headlineoverall_reo_recovery_pct, concentrate_reo_grade_wt_pct
Plant-data accounting view of the Mountain Pass bastnaesite flotation circuit: the same 12-cell rougher -> cleaner 1 -> cleaner 2 -> cleaners 3 and 4 -> concentrate, with cleaner-tail recycles and a scavenger plus regrind returning to the rougher. Instead of kinetics, each bank's recovery of each component is a fixed fraction calibrated to published plant data (Pradip and Fuerstenau 2013, the Table 1 product fit), so the model is a linear split balance with recycles. Use it for what-if questions such as 'what if the rougher recovered 98% of the rare earths'. It does not respond to throughput or equipment size (use mountain_pass_flotation for that). Flows are reported at the 1700 t/d plant basis. The headline KPIs are overall REO recovery and concentrate REO grade; rare earths are one lumped oxide.
paramsrougher_reo_recovery_frac, scavenger_reo_recovery_frac, cleaners34_reo_recovery_frac, rougher_inert_gangue_recovery_frac, flotation_feed_reo_grade_wt_pct
kpisoverall_reo_recovery_pct, concentrate_reo_grade_wt_pct, concentrate_mass_pull_wt_pct, concentrate_reo_kg_per_hr, concentrate_dry_solids_kg_per_hr, tailings_reo_grade_wt_pct … (+5)
metalsCe, La, Nd, Pr
ore_crusher
headlinespecific_crushing_energy_kwh_per_t, crushing_power_kw
Size reduction of run-of-mine ore or coal refuse in one crusher stage: feed at a given 80 % passing size (F80) -> crusher -> product at a target P80, with Rosin-Rammler size distributions. Power follows Bond's law, 10 x throughput x Bond work index x (1/sqrt(P80) - 1/sqrt(F80)), so energy per tonne rises with ore hardness and with how fine the product must be. The model also names the crusher class (primary jaw or gyratory, secondary cone or roll, tertiary short-head cone, roll or hammer mill) that fits the product size. The model is ore-agnostic: ore hardness enters only through the work index, so it tracks no metal. The upstream reference is 2 t/h of REE-bearing coal refuse at a work index of 12 kWh/t.
paramscrusher_product_p80_mm, bond_work_index_kwh_per_t, crusher_feed_p80_mm, crusher_feed_t_per_hr
kpisspecific_crushing_energy_kwh_per_t, crushing_power_kw, feed_p80_mm, product_p80_mm, reduction_ratio, recommended_crusher_stage
ree_cost_screening
headlinetotal_operating_cost_central_usd_per_kg, total_operating_cost_low_usd_per_kg, total_operating_cost_high_usd_per_kg, total_capital_cost_central_usd_per_kg
Screening estimate, not a flowsheet solve: deposit size (feedstock tonnes) and rare earth grade -> contained rare earths -> production-cost bands in USD 2021 per kg of rare earth product. The bands are the scaling correlations of Fritz, Tarka and Mauter, 'Assessing the economic viability of unconventional rare earth element feedstocks', Nature Sustainability (2023), doi:10.1038/s41893-023-01145-1, as implemented in PROMMIS prommis.costing.ree_costing (calculate_REE_costing_bounds). Each band is (a +/- da) x (contained rare earths in tonnes)^(k +/- dk), so only grade x tonnage matters and bigger, richer deposits screen cheaper. Bands cover total capital, total operating, mining, beneficiation, chemical extraction, enrichment and separation, and two combined stages; the headline is the central total operating cost from the nominal coefficients. Grade and cost are on the source's basis; PROMMIS does not say whether that is elemental REE or oxide (TREO). Contained rare earths are capped at 2e8 t; far from the reference deposit (about 26,090 t contained) the bands are power-law extrapolations. Use the ranges to compare deposits at order-of-magnitude level, not as a deposit valuation.
paramsdeposit_feedstock_tonnes, deposit_ree_grade_pct
kpistotal_operating_cost_central_usd_per_kg, total_operating_cost_low_usd_per_kg, total_operating_cost_high_usd_per_kg, total_capital_cost_central_usd_per_kg, total_capital_cost_low_usd_per_kg, total_capital_cost_high_usd_per_kg … (+13)
metalsSc, Y, La, Ce, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb, Lu
ree_ion_exchange
headlineree_recovery_pct, service_time_h, capital_cost_usd
A fixed-bed ion-exchange column of S950 resin captures six rare earths (La, Dy, Ho, Er, Yb, Sm) from dilute treated acid mine water (Hermassi et al. 2022 lab data: about 1 mg/L La at 1 mL/min through a 1 m x 15 mm bed). Feed -> resin bed -> effluent, with the breakthrough curve from a Clark model with Freundlich parameters fitted to that data, and a WaterTAP-style cost model for vessel, resin, backwash and resin replacement or NaCl regeneration. The headline KPI is the share of the fed rare earths held on the resin over one loading cycle. It is lab scale, so costs are illustrative and revenue is cents per year; it pre-concentrates REEs ahead of solvent extraction or precipitation.
paramsix_feed_flow_ml_per_min, ix_bed_depth_m, ix_bed_diameter_m, ix_breakthrough_endpoint_frac, ix_feed_ree_conc_factor, ix_target_ree … (+2)
kpisree_recovery_pct, service_time_h, la_recovery_pct, dy_recovery_pct, ho_recovery_pct, er_recovery_pct … (+13)
metalsLa, Dy, Ho, Er, Yb, Sm
ree_leaching
headlinetotal_ree_recovery_pct, nd2o3_recovery_pct, fe2o3_co_leach_pct
Co-current slurry leaching of West Kentucky No. 13 coal refuse, the rare-earth feed of the University of Kentucky pilot plant: coal-refuse solids and acid liquor -> 1 to 3 stirred leach tanks in series -> pregnant leach solution (to solvent extraction) + leached residue. Rates depend on free H+, so acid strength, liquid-to-solid ratio, tank count and tank volume trade rare-earth recovery against co-leaching of Al, Fe and Ca gangue. The acid can be sulfuric (upstream default) or, through PROMMIS's mixed-acid property package, HCl alone or mixed with H2SO4; any HCl selects that package. Recoveries are per oxide (Sc, Y, La, Ce, Pr, Nd, Sm, Gd, Dy). Points that spend the acid (any tank above pH 6) are refused: the model has no neutral-liquor chemistry.
paramsree_leach_h2so4_mol_per_l, ree_leach_hcl_mol_per_l, ree_leach_tank_count, ree_leach_tank_volume_gal, ree_leach_liquid_l_per_hr, ree_leach_solid_kg_per_hr
kpistotal_ree_recovery_pct, sc2o3_recovery_pct, y2o3_recovery_pct, la2o3_recovery_pct, ce2o3_recovery_pct, pr2o3_recovery_pct … (+11)
metalsSc, Y, La, Ce, Pr, Nd, Sm, Gd, Dy, Al, Fe, Ca
ree_magnet_recycling_superstructure
headlinenet_present_value_usd, breakeven_nd_price_usd_per_kg, total_plant_cost_usd, annual_operating_cost_usd_per_yr
Techno-economic pathway selection for recycling NdFeB magnets from end-of-life products (PROMMIS superstructure, run on its 3-stage upstream example). Collected products -> stage 1 disassembly in whole units -> stage 2 processing (2 options) -> stage 3 Nd recovery (3 options, Fe rejected). Each option carries a recovery fraction, operating costs, a piecewise equipment cost curve and an operator count; QGESS and Keim (2019) factors turn the chosen route into capital, operating cost and yearly cash flow (2024 build, 2025-2028 operation). Every request is an optimization, so solve, sweep and optimize all return the best pathway and its economics. The costs are illustrative PROMMIS example costs, not a published plant study, and the options are named only as (stage, option). The upstream model's fixed caps on flows, operators and costs are widened, and yearly sales and the cost lines tied to them may go negative (byproduct disposal costs at Nd values under about 0.19 USD/kg), so neither excludes a pathway. At the defaults it picks (1, 1) -> (2, 2) -> (3, 3) with an NPV of -16.44 M USD and a breakeven Nd price of 261.46 USD/kg, far above the 45.43 USD/kg product value.
paramsrecycled_nd_value_usd_per_kg, eol_magnet_collection_rate_frac, recycling_operator_wage_usd_per_yr, recycling_discount_rate_frac, disassembly_throughput_products_per_yr, recycling_plant_lifetime_yr … (+2)
kpisnet_present_value_usd, breakeven_nd_price_usd_per_kg, breakeven_nd_value_usd_per_kg, total_plant_cost_usd, annual_operating_cost_usd_per_yr, annual_revenue_usd_per_yr … (+8)
metalsNd, Fe
ree_oxalate_precipitation
headlineree_precipitation_recovery_pct, ree_oxalate_purity_pct
Precipitation of mixed rare earth oxalates from a leach or strip liquor: liquor carrying nine rare earths (Sc, Y, La, Ce, Pr, Nd, Sm, Gd, Dy) plus Al, Ca and Fe -> oxalic acid dosing in one stirred tank -> rare earth oxalate cake and a barren liquor. Each element's conversion follows an isotherm fitted to MINTEQ equilibrium runs, exp(-(E / dose) ** N) with the dose in g/L, so the model predicts how much of each rare earth precipitates, how much iron and aluminum come down with it, and the purity of the cake. Light rare earths precipitate readily, scandium poorly. Calcium is carried through but its conversion is pinned at a negligible minimum, so calcium oxalate co-precipitation is not modeled. The cake is the feed to calcination into mixed rare earth oxide.
paramsoxalic_acid_dose_mg_per_l, precip_liquor_flow_l_per_hr, precip_ree_each_mg_per_l, precip_fe_mg_per_l, precip_al_mg_per_l, precip_ca_mg_per_l … (+1)
kpisree_precipitation_recovery_pct, ree_oxalate_purity_pct, precipitate_kg_per_hr, scandium_recovery_pct, iron_coprecipitation_pct, aluminum_coprecipitation_pct … (+2)
metalsAl, Ca, Fe, Sc, Y, La, Ce, Pr, Nd, Sm, Gd, Dy
ree_roaster
headlineleachable_ree_pct, roaster_heat_duty_mw
Thermal pretreatment of rare earth bearing coal refuse ahead of acid leaching: refuse (carbon, pyrite, kaolinite, quartz, limestone and nine rare earths) -> roaster bed of fixed volume and voidage, swept by hot flue gas -> calcine to leaching plus an off-gas. Carbon, hydrogen, organic sulfur and pyrite burn to completion (stoichiometric) and kaolinite dehydrates fully; kinetic rate laws calcine the limestone and convert acid-insoluble rare earth phases into leachable oxides. The model predicts how much of the rare earth content a leach can reach for a given bed temperature, bed volume (residence time) and gas supply; off-gas SO2 follows the feed and gas rates, not the temperature. A feed the gas cannot burn (about 24.1 mol O2 per kg of refuse) is refused as oxygen-starved. The upstream reference roasts 1030 kg/h of West Kentucky No. 13 coal refuse in a 4 m3 bed; leachability peaks near 600 C.
paramsroast_temperature_k, roaster_bed_volume_m3, roaster_flue_gas_mol_per_s, roaster_refuse_feed_kg_per_hr, roaster_flue_gas_temp_k
kpisleachable_ree_pct, roaster_heat_duty_mw, calcine_kg_per_hr, calcine_leachable_reo_ppm, limestone_calcination_pct, offgas_o2_mol_pct … (+1)
metalsSc, Y, La, Ce, Pr, Nd, Sm, Gd, Dy
ree_solvent_extraction
headlinetotal_ree_extraction_pct, dy_extraction_pct, nd_extraction_pct
Counter-current solvent extraction (SX) loading rare earths from an acidic coal-refuse leachate into DEHPA in kerosene, as in the University of Kentucky pilot plant: leachate + barren organic -> 1 to 20 counter-current equilibrium stages -> loaded organic (to stripping) + raffinate. Rare-earth distribution follows an empirical fit in pH and DEHPA dosage, so the O:A ratio, dosage, stage count and feed pH trade heavy-REE (Y, Dy) capture, light-REE (La, Ce, Nd) capture and Fe/Al/Ca co-extraction. Sc, Al, Ca and Fe use constant distribution coefficients. The feed enters at bisulfate equilibrium, so the set pH is the pH the first stage sees. Any HCl added to the leachate builds on PROMMIS's mixed-acid property package; it acts only through pH.
paramsree_sx_organic_to_aqueous_ratio, ree_sx_dehpa_dosage_vol_pct, ree_sx_stage_count, ree_sx_leachate_ph, ree_sx_aqueous_feed_l_per_hr, ree_sx_hcl_addition_mol_per_l
kpistotal_ree_extraction_pct, sc_extraction_pct, y_extraction_pct, la_extraction_pct, ce_extraction_pct, pr_extraction_pct … (+12)
metalsSc, Y, La, Ce, Pr, Nd, Sm, Gd, Dy, Al, Ca, Fe
uky
headlineoverall_ree_recovery_pct, product_purity_pct
REE recovery from coal byproducts. Leaching → solvent extraction (rougher + cleaner) → precipitation → roasting.
paramsleach_liquid_feed_rate_l_per_hr, leach_liquid_feed_h_mg_per_l, leach_liquid_feed_so4_mg_per_l, leach_solid_feed_mass_kg_per_hr, roaster_gas_inlet_temp_k, roaster_gas_inlet_pressure_pa … (+1)
kpisree_product_kg_per_hr, ree_feed_kg_per_hr, overall_ree_recovery_pct, product_purity_pct, overall_metal_recovery_pct, leaching_metal_recovery_pct … (+3)
metalsSc, Y, La, Ce, Pr, Nd, Sm, Gd, Dy, Al, Fe, Ca