What a Compositional Flash Calculation Solves

A compositional flash calculation asks how a specified feed distributes between the phases allowed by a thermodynamic model. In a pressure-temperature, or PT, flash, you supply overall mole fractions, absolute pressure and temperature. The solver determines a supported equilibrium state, including a liquid-vapor split when the model predicts two phases. The phase compositions are outputs, not extra feed specifications.

This lesson follows one reproducible C1–nC6 Peng–Robinson case at 280 K and 2 MPa absolute. We replayed the site's pinned CoolProp 8.0.0 implementation and retained the full-precision phase compositions for the ledger. The explorer selects a component and scales the feed amount; it does not re-solve equilibrium when you move the amount slider. Use the linked flash workflow to change temperature, pressure or composition.

The overall feed composition z describes component moles divided by total feed moles. Here z is 0.50 methane, 0.15 ethane, 0.15 propane, 0.10 normal butane, 0.05 normal pentane and 0.05 normal hexane. The entries sum to one. On a 100-mol feed basis, they correspond to 50, 15, 15, 10, 5 and 5 mol respectively.

The component list matters as much as the numbers. Normal butane and isobutane are different substances, and a heavy petroleum fraction is not automatically equivalent to normal hexane. This particular workflow supports the six listed hydrocarbons with zero binary interaction parameters. Its software bounds are 180–450 K and 0.1–10 MPa absolute; those bounds describe accepted inputs, not an assurance of accuracy throughout that region.

Liquid composition x describes moles of a component divided by all liquid-phase moles. Vapor composition y uses all vapor-phase moles as its denominator. Each phase composition vector separately sums to one. The vapor mole fraction beta instead describes total vapor moles divided by total feed moles; 1 − beta is the liquid mole fraction of the feed.

For this case beta is 0.657564116, giving 65.756412 mol of vapor and 34.243588 mol of liquid per 100 mol of feed. Methane's vapor-phase composition is about 0.70979194, but that does not mean 70.98% of the whole feed becomes methane vapor. Multiply the vapor amount by its methane composition to find that component's actual vapor allocation.

Every component that enters must be present in the returned phases because this PT case contains no reaction or material loss. Dividing Fzi = Lxi + Vyi by the feed amount gives zi = (1 − beta)xi + beta yi. The liquid and vapor allocation terms share the feed-mole denominator, so they can be added directly. The compositions xi and yi by themselves cannot simply be added.

The table retains the same fixed state for all six components. Composition columns use moles per mole of their own phase; allocation columns use moles per mole of feed. Display values are rounded, while the balance checks use the full-precision snapshot. The largest unrounded component residual is approximately 5.6 × 10⁻¹⁷ mol/mol feed, so small apparent rounding differences in the displayed rows are not missing material.

A mole counts an amount of substance, while mass depends on each component's molecular mass. This vapor is enriched in lighter molecules relative to the liquid. Its fraction of total moles is therefore larger than its fraction of total mass: approximately 65.7564% of the feed moles are vapor, compared with 44.8913% of the feed mass. Both numbers can be correct for the same equilibrium split.

An in-situ phase volume fraction requires phase volumes at the stated conditions. It is not beta, and it is not a flow holdup measurement. To calculate a volume fraction you would need consistent phase amounts and molar volumes, or masses and densities. Keep the basis in every output label when carrying flash results into separator, storage or flow calculations; changing the denominator changes the quantity being reported.

A component ledger alone permits many mathematically balanced splits. Equilibrium adds the requirement that each present component has equal fugacity in the liquid and vapor phases at the common pressure and temperature. The equation of state supplies the phase-dependent fugacity coefficients. The compositions affect those coefficients, so phase composition and phase amount must be solved together rather than assigned independently.

The workflow uses the pinned library's PT stability search and phase-split calculation, then audits its returned phase pressures, composition sums, component balances and fugacity agreement. For this snapshot the maximum log-fugacity mismatch is about 1.9 × 10⁻¹⁰ and the relative EOS pressure residual about 8.8 × 10⁻¹². These are numerical consistency diagnostics for this model; they do not measure agreement with an actual fluid sample.

The teaching example uses Peng–Robinson mixing rules with every binary interaction parameter kij set to zero. That is an explicit model choice rather than a sample-specific interaction fit. The linked C1–nC6 workflow does not include water, sour gases, heavy pseudocomponents, wax or an additional liquid phase. Those phenomena need a model and component specification that actually support them.

A returned single phase also needs careful reading: no absent-phase composition or arbitrary vapor fraction should be invented to fill a two-phase table. Near a vanishing phase or critical region, the workflow applies conservative separation and conditioning gates. Use the PT flash to study a specified state and the phase-envelope tool for its separately supported boundary task. Changing total feed moles in this ledger only rescales amounts at the same fixed equilibrium state.