The 3D tunnel version treats the catalytic bed as a black box: you set an "abatement efficiency" slider directly and the plant obeys it, no matter how much gas you push through. Real packed-bed reactors do not work that way — a bed only has time to convert what a molecule's residence time allows. This 2D companion removes the efficiency slider and derives it instead from real first-order surface-reaction kinetics and the ideal-gas volumetric flow through the bed:
NCG flow = steam flow × NCG fraction
H2S flow = NCG flow × H2S share
n_dot = NCG flow / M_avg (mol/s, M_avg ≈ 0.044 kg/mol, CO2-dominated)
Q_gas = n_dot·R·T / P (m³/s, ideal gas at bed conditions, T=400K)
A_bed = π·(d_bed/2)²
v_bed = Q_gas / A_bed (m/s, superficial gas velocity)
τ = L_bed / v_bed (s, residence time; L_bed = 2.2 m, fixed)
η = 1 − exp(−k·τ) (first-order conversion — this IS the abatement efficiency)
H2S emitted = H2S flow × (1 − η)
S captured = 0.94 × H2S flow × η (mass ratio S/H2S)
Parasitic steam = NCG flow × 3 (motive steam per kg NCG, ejector train)
Parasitic power = parasitic steam × 700 kJ/kg / 1000
Net power = steam flow × 700 kJ/kg / 1000 − parasitic power
- The coupling the 3D model misses: push more steam through the plant and NCG flow rises, so the volumetric gas flow through the bed rises too — the same bed now passes gas faster, residence time τ falls, and η drops even though the catalyst itself hasn't changed. Try dragging steam flow up and watch η fall in real time.
- Bed vessel diameter — a wider bed slows the superficial velocity for the same gas flow (v = Q/A), buying more residence time and higher η at the cost of a bigger, pricier vessel.
- Catalyst rate constant k — an intrinsic property of the catalyst and its operating temperature; a hotter, fresher, more active bed converts H₂S faster per second it sees the gas.
- Each yellow parcel entering the bed is sampled from a real exponential capture-time distribution: draw T = −ln(U)/k for a random U; if T ≤ τ the parcel is captured at fraction T/τ of the bed's length (visualized as the parcel darkening partway through the bed), otherwise it survives the whole bed and rides on to the stack as plume. Averaged over many parcels this reproduces η = 1 − e−kτ exactly — the same law used for the readouts above.