An Error Budget for the SRK and PR Equations of State: α-Function Parameterization and Volume Translation in a Reproducible Pure-Fluid Saturation-Property Benchmark

The Soave–Redlich–Kwong (SRK) and Peng–Robinson (PR) equations of state remain the default process-simulation models, yet their errors are usually quoted qualitatively. Both are benchmarked here in strictly predictive mode (inputs Tc, pc, ω) against reference Helmholtz-energy equations of state for 37 fluids in five chemical families: 1480 saturation states at reduced temperatures from 0.50 to 0.985, four properties (vapor pressure, enthalpy of vaporization, saturated-liquid, and saturated-vapor density), fluid-level bootstrap confidence intervals, and multiplicity-adjusted paired tests. Unlike previous assessments of saturation properties, six generalized α-functions (Soave, its 2019 update, Twu-95, the consistent generalized Twu function, Coquelet, Mahmoodi–Sedigh) are exchanged across both cubics; component-specific Twu parameters serve as an out-of-sample fitted reference, the α-function of each cubic is refitted to the same data so that the SRK–PR difference is measured at matched parameterization, and a volume-translation ladder is applied to both cubics. The α-function sets the vapor-pressure error: the classical SRK–PR vapor-pressure gap of 0.68 percentage points shrinks to 0.0–0.4 points under the other generalized α-functions and, with α-functions fitted to the same data, is 0.58 points for the one-parameter Soave form but 0.15 for a three-parameter form. The enthalpy of vaporization also carries a volumetric error from a too-ideal cubic saturated vapor. Water, methanol, and ethanol are the only fluids that no generalized α-function describes to within 2%; fitted parameters show this to be mainly a failure of generalization and of the Soave form, not of the cubic. Saturated-liquid densities follow a separate budget set by the cubic form, the translation constant, and the critical-compressibility mismatch; translated PR beats translated SRK for every fluid. Fitted α-parameters are tied to the critical constants used to fit them. The pipeline is open source.