ePC-SAFT: an electrolyte equation-of-state engine with exact derivatives for phase and reaction equilibrium and parameter estimation

Electrolyte parameter estimation and equation-oriented process models require properties of a solved phase and reaction equilibrium and their responses to temperature, pressure, feed and molecular parameters. The ePC-SAFT package supplies these calculations in native C++ through a Python interface. Published hard-chain, dispersion, association, polar and electrolyte contributions define a selected Helmholtz energy; association and pressure closure, constrained equilibrium and exact derivatives share its mathematical implementation. One equilibrium driver handles declared neutral, charged and reacting phase specifications, while explicit tangent-plane searches test additional phases. Implicit derivatives of the defining equations supply observation Jacobians, least-squares and guarded minimax parameter estimation, and first and second derivatives for process models. Comparisons span pure-fluid saturation, neutral and associating mixtures, aqueous and mixed electrolytes, solvent transfer, liquid splits and reacting systems, with calibration, withheld prediction and published-model reproduction identified separately. With fixed published parameters, propane saturation pressures and liquid densities deviate from measurements by 2.70 % and 0.305 % on average; on 19 temperatures withheld from a threetemperature refit, the refit gives 1.00 % and 0.180 %, against 1.07 % and 0.228 % for the published parameters. The comparisons also expose failed mixed-salt transfer, large reactive-amine pressure discrepancies and unresolved source conventions. Numerical verification gives liquidliquid mole fractions within 4.62 × 10−9 of an independent matched electrolyte implementation and selected solved-state derivatives within 3.40 × 10−10 relative. Settler and hydrogen-mixture examples show how the same calculations enter engineering models, with physical accuracy bounded by the stated parameter domains and data.