Thermodynamic models¶

Convenience builders that turn component names into a ready EOS or gamma-phi EquilibriumModel (Peng-Robinson, NRTL, UNIQUAC, UNIFAC), plus the lightweight modified-Raoult helpers for quick ideal-ish estimates.

Model builders¶

models ¶

Model bridge: turn component names + a method choice into an equilibrium model.

The thermo equilibrium models (EOSModel and GammaPhiModel) take array constants (tc, pc, omega) and, for gamma-phi, an activity model. A flowsheet, however, works in component names. This module resolves names to those arrays (reusing the cached lookup in fugacio.sim.properties) and assembles the activity model from the curated binary database (NRTL / UNIQUAC) or predictive group contribution (UNIFAC / modified UNIFAC), returning a ready, differentiable EquilibriumModel.

The returned object is what the gamma-phi-aware unit operations (fugacio.sim.separations) and the T-x-y / P-x-y / azeotrope helpers (fugacio.sim.diagrams) consume, so a flowsheet can switch from Peng-Robinson to NRTL by swapping one constructor call, and stays end-to-end differentiable, including with respect to the activity-model parameters.

Classes:

Name	Description
`UnifacModel`	An `ActivityModel` adapter over predictive UNIFAC.

Functions:

Name	Description
`eos_model_for`	Build an `EOSModel` for named `components`.
`nrtl_model_for`	Gamma-phi model with NRTL liquid from the curated binary database.
`uniquac_model_for`	Gamma-phi model with UNIQUAC liquid from the curated database (with `r`/`q`).
`unifac_model_for`	Gamma-phi model with a predictive UNIFAC liquid (no fitted parameters needed).

UnifacModel `dataclass` ¶

UnifacModel(components: tuple[str, ...], dortmund: bool)

An ActivityModel adapter over predictive UNIFAC.

Wraps fugacio.thermo.unifac_activity (classic, Hansen VLE parameters) or fugacio.thermo.modified_unifac_activity (Dortmund, T-dependent) so that group-contribution predictions present the same ln_gamma(x, T) API as the fitted activity models. Carries no fitted leaves: it is a pure predictor keyed by the (static) component names.

Methods:

Name	Description
`ln_gamma`	Log activity coefficients predicted by (modified) UNIFAC.

ln_gamma ¶

ln_gamma(x: Array, t: ArrayLike) -> Array

Log activity coefficients predicted by (modified) UNIFAC.

eos_model_for ¶

eos_model_for(
    components: Sequence[str],
    *,
    eos: CubicEOS = PR,
    kij: Array | None = None,
    use_database_kij: bool = False,
) -> EOSModel

Build an EOSModel for named components.

Pass use_database_kij=True to fill the binary interaction matrix from the curated ChemSep Peng-Robinson k_ij set (fugacio.thermo.kij_from_database); pairs without a curated value stay at zero. An explicit kij takes precedence.

nrtl_model_for ¶

nrtl_model_for(
    components: Sequence[str],
    *,
    eos: CubicEOS = PR,
    kij: Array | None = None,
    vapor: str = "ideal",
    poynting: bool = False,
    phi_saturation: bool = False,
    strict: bool = False,
    alpha_default: float = 0.3,
) -> GammaPhiModel

Gamma-phi model with NRTL liquid from the curated binary database.

Pairs absent from the database default to athermal interaction unless strict=True; see fugacio.thermo.nrtl_from_database.

uniquac_model_for ¶

uniquac_model_for(
    components: Sequence[str],
    *,
    eos: CubicEOS = PR,
    kij: Array | None = None,
    vapor: str = "ideal",
    poynting: bool = False,
    phi_saturation: bool = False,
    strict: bool = False,
) -> GammaPhiModel

Gamma-phi model with UNIQUAC liquid from the curated database (with r/q).

unifac_model_for ¶

unifac_model_for(
    components: Sequence[str],
    *,
    dortmund: bool = False,
    eos: CubicEOS = PR,
    kij: Array | None = None,
    vapor: str = "ideal",
    poynting: bool = False,
    phi_saturation: bool = False,
) -> GammaPhiModel

Gamma-phi model with a predictive UNIFAC liquid (no fitted parameters needed).

Set dortmund=True for modified UNIFAC (Dortmund) with temperature-dependent group interactions; otherwise classic UNIFAC is used.

Modified-Raoult helpers¶

vle ¶

Vapor-liquid equilibrium for binary mixtures (modified Raoult's law).

A deliberately small but real vertical slice that sits on top of fugacio.thermo: it combines Antoine vapor pressures with Margules activity coefficients to compute a bubble-point pressure. Everything is differentiable end-to-end with respect to composition, temperature, and the activity-model parameters.

Functions:

Name	Description
`antoine_psat`	Saturation pressure from the base-10 Antoine equation.
`bubble_pressure`	Bubble-point pressure and vapor composition for a binary at fixed `T`, `x`.

antoine_psat ¶

antoine_psat(
    temperature: ArrayLike,
    a: ArrayLike,
    b: ArrayLike,
    c: ArrayLike,
) -> Array

Saturation pressure from the base-10 Antoine equation.

log10(Psat) = a - b / (temperature + c)

Pressure and temperature units follow whatever convention the Antoine constants were fit in; Fugacio does not impose one here.

bubble_pressure ¶

bubble_pressure(
    x1: ArrayLike,
    temperature: ArrayLike,
    antoine1: Antoine,
    antoine2: Antoine,
    a12: ArrayLike = 0.0,
    a21: ArrayLike = 0.0,
) -> tuple[Array, Array]

Bubble-point pressure and vapor composition for a binary at fixed T, x.

Uses modified Raoult's law p_i = x_i * gamma_i * Psat_i with two-parameter Margules activity coefficients. Setting a12 = a21 = 0 recovers ideal (Raoult's law) behaviour.

Returns:

Type	Description
`tuple[Array, Array]`	`(pressure, y1)`, where `y1` is the vapor mole fraction of component 1.