qfoldit-mining

qfoldit-mining

Kinetic modeling skill for three related processes in refractory ore processing and effluent treatment:

1. Bio-oxidation (bioox_kinetics.py) -- microbial oxidation of sulfide minerals (arsenopyrite/pyrite) that locks up gold, freeing it for downstream leaching. Model: Shrinking Core Model (reaction-controlled) x Arrhenius x microbial cardinal-temperature response x pH response.
2. Biosorption (biosorption_kinetics.py) -- recovery of dissolved metal (gold-cyanide complex, copper, REE, lithium) onto biomass. Model: Langmuir/Freundlich equilibrium x pseudo-second-order kinetics x pH-dependent activity (mechanism differs for cationic metals vs anionic complexes like Au(CN)2-).
3. Cyanide biodegradation (cyanide_kinetics.py) -- biological detoxification of free/WAD cyanide in tailings or mill effluent by cyanide-degrading bacteria. Model: Aiba substrate-inhibition kinetics (cyanide is BOTH the growth substrate and, at high concentration, toxic to the degrading culture -- rate is non-monotonic in concentration, not simple Monod).

Read references/model_documentation.md before answering -- it contains the literature basis, all equations, calibration status, and critical caveats (especially: default rate/capacity constants are generic literature-range placeholders, NOT validated for any specific qFoldIT ore or site, until fitted against real assay data).

How to handle a request

1. Identify which process the user is asking about (bio-oxidation of ore, biosorption of dissolved metal, or cyanide degradation in effluent/tailings) and which metal/system, if biosorption (gold_cyanide, copper_cationic, ree_trivalent, lithium_cationic, or a custom parameter set).
2. Validate inputs: pH should be a plausible acidic bioleaching range (typically 0.5-6); temperature in a plausible mesophilic range (typically 10-50 C) -- flag values outside this as unusual and ask for confirmation rather than silently extrapolating. Particle size, concentration (Ce) must be positive. For cyanide degradation, flag concentrations above roughly 1000-1500 mg/L as entering a regime where the source study's fit is less certain (sparse literature data that high) -- see references.
3. Run the model via scripts/bioox_kinetics.py, scripts/biosorption_kinetics.py, or scripts/cyanide_kinetics.py (see references for function signatures and examples).
4. Always report calibration status alongside any numeric forecast: state plainly whether the constants used are the generic literature defaults or have been fitted to real data the user supplied. Never present a placeholder-calibrated forecast as a validated production guarantee.
5. If the user has real assay/pilot data (time-series conversion data, equilibrium isotherm data, kinetic uptake data, or rate-vs- concentration data for cyanide degradation), use fit_A_from_data, fit_langmuir, fit_freundlich, fit_pseudo_second_order, or fit_aiba to recalibrate before forecasting -- this is always preferable to the defaults.
6. Reactor mode matters for bio-oxidation: ask (or confirm) whether the reactor is continuous/mature-culture (e.g. BIOX-style CSTR train) or batch/fresh-inoculation (e.g. heap start-up) -- these give drastically different time-to-target-conversion.
7. For cyanide degradation, never give an absolute treatment time (hours/days to reach a target residual concentration) without first flagging the X0 (starting biomass) caveat -- the literature rate constants describe specific (per-unit-biomass) rates, and the translation to an absolute timeline is extremely sensitive to a biomass loading value this skill cannot supply on its own (verified: a 400x range of X0 changes a 300-hour outcome from "barely moved" to "cut by nearly half" using the same rate constants). Absolute simulate_batch_treatment/time_to_target_residual results should be presented as illustrating the SHAPE of the response, not a commissioning timeline, unless the user has supplied a real measured biomass density.