Gypsum (CaSO₄·2H₂O) loses its crystallization water in two sequential, first-order steps as it's heated — the classic A→B→C scheme:
CaSO4·2H2O -> CaSO4·0.5H2O + 1.5 H2O (k1)
CaSO4·0.5H2O -> CaSO4 + 0.5 H2O (k2)
df_di/dt = -k1 * f_di
df_hemi/dt = k1 * f_di - k2 * f_hemi
df_anh/dt = k2 * f_hemi
k(T) = A * exp(-Ea / (R * T_K)) (Arrhenius)
Both rate constants follow the Arrhenius law with literature-scale activation energies (Ea₁ ≈ 60 kJ/mol for the dihydrate→hemihydrate step, Ea₂ ≈ 100 kJ/mol for hemihydrate→anhydrite — the second, deeper dehydration needs more thermal energy). The pre-exponential factors are tuned so the process finishes over tens of simulated seconds instead of the real industrial minutes-to-hours, so the kinetics stay interactive. Smaller particles dehydrate faster (less diffusion path for water vapor to escape), which the particle-size slider models as a 1/d scaling on both rate constants. The kiln furnace holds the target temperature you set, ramping toward it at the chosen heating rate; each grain in the bed converts phase individually once the macroscopic conversion fraction reaches it, so you see the bed visibly shift color from blue-grey (dihydrate) through tan (hemihydrate/plaster of Paris) to pale grey (anhydrite) as calcination proceeds.