Checking a Flash Result: Component Balances, Composition Sums, and Phase Fractions

A flash calculation reports a small set of numbers describing how a feed splits into equilibrium vapor and liquid: a vapor fraction β, a liquid composition x_i, and a vapor composition y_i. Because these come from an iterative solver, they should be verified before use. A trustworthy result passes three independent families of checks: each phase’s mole fractions sum to one on their own, the component balance closes for every component, and β satisfies the Rachford–Rice equation with K-values consistent with y_i = K_i x_i. This article sets out those checks in a reusable form, works a small binary example with a component-by-component allocation table, and separates a phase fraction from a phase composition — the most common point of confusion when reading a flash result.

What a flash result contains

A flash calculation takes a feed of F total moles with composition z_i and splits it into a vapor and a liquid in vapor–liquid equilibrium (VLE) — the condition in which each component’s chemical potential is equal in both phases, so the phases coexist without net transfer. The most common specification is the isothermal, isobaric (P–T) flash: temperature, pressure, and feed composition are given, and x_i, y_i, and β are solved for, as in the standard formulation of the two-phase flash problem. Other common specifications are bubble-point and dew-point calculations at fixed temperature or pressure, and the pressure–enthalpy (P–H) flash used, for example, for the separator downstream of a valve, which additionally requires an energy balance. These variants and their typical numerical treatment are cataloged in chapter 7.5 of Skogestad’s textbook material on entropy and equilibrium.

The outputs mean different things. For a feed of F moles, the vapor holds βF moles and the liquid (1 − β)F moles. Component i appears as βF y_i moles in the vapor and (1 − β)F x_i moles in the liquid. Each phase’s composition — x_i in the liquid, y_i in the vapor — sums to one independently, and neither x_i nor y_i is multiplied by the feed size until a component inventory is actually requested. β is a phase fraction, an amount relative to the total feed; x_i and y_i are compositions, amounts within a particular phase.

The material balance behind a valid result

Every two-phase flash result must satisfy the component balance z_i = (1 − β)x_i + β y_i. With the equilibrium ratio, or K-value, defined as K_i = y_i/x_i, the phase compositions are fixed once β is chosen: x_i = z_i / (1 + β(K_i − 1)) and y_i = K_i x_i. Requiring both phase compositions to sum to one collapses the balance into the Rachford–Rice equation,

F(β) = Σ_i z_i(K_i − 1) / (1 + β(K_i − 1)) = 0,

published by Rachford and Rice in 1952. Its admissible root must keep all denominators and both phase compositions positive. This is the material-balance core solved in flash tools such as the torch-flash two-phase flash documentation, which also couples the balance to thermodynamics.

For real fluids the K-values are not constants. An equation of state — a model such as the Peng–Robinson equation — supplies a fugacity coefficient φ_i for each phase, a dimensionless correction for non-ideal behavior. The equilibrium condition of equal chemical potential then reduces to φ_i^V y_i = φ_i^L x_i, so K_i = φ_i^L / φ_i^V, an expression that also covers supercritical components. Because each fugacity coefficient depends on its own phase composition, the Rachford–Rice balance and the fugacity equations must be iterated together; implementations may add a tangent-plane stability analysis to decide whether a two-phase split exists at all, and typically start from explicit or Wilson-type initial K-values.

Five checks that a flash result must pass

1. Both phase compositions sum to one, independently. Compute Σ x_i and Σ y_i as two separate numbers, because the phases can have different molecular weights and compositions. A sum of one for the liquid does not ensure the same for the vapor if the K-values or the final iteration are inconsistent.
2. The component balance closes for every component. Check βF y_i + (1 − β)F x_i = F z_i component by component. A convenient device is a component allocation ledger, which lists the feed, liquid, and vapor contributions of each component side by side so that any misallocation is visible at a glance.
3. The K-values are consistent with the compositions. Dividing each reported y_i by the corresponding x_i should reproduce the K_i used in the solve; in an equation-of-state run, the converged compositions should also reproduce K_i = φ_i^L/φ_i^V. A mismatch signals that compositions and K-values came from different iterations.
4. The Rachford–Rice residual is zero at the reported β. Substituting the reported β and K_i into the equation above should give approximately zero, within the solver’s convergence tolerance, and the root must be admissible: all denominators 1 + β(K_i − 1) and all compositions positive.
5. The specification and phase state make sense. β must lie between zero and one; a result pinned at a bound, or a trivial solution with x_i = y_i = z_i, suggests the feed is actually single-phase at the specified conditions and a stability analysis is needed. For a P–H flash, close the energy balance as well as the material balance.

Worked example: checking a synthetic binary flash

The following hypothetical example applies the checks to a teaching case from a lab walkthrough for ideal VLE and the P–T flash: an ideal-solution binary on a 1 mol feed basis, with components A (lighter) and B (heavier), z_A = z_B = 0.5, at T = 350 K and a flash pressure of 95 kPa. The supplied property records give synthetic saturation pressures of P_A^sat = 150 kPa and P_B^sat = 60 kPa at 350 K; these are defined teaching data, not measurements of a real fluid.

For an ideal solution following Raoult’s law, K_i = P_i^sat / P, so K_A = 150/95 ≈ 1.579 and K_B = 60/95 ≈ 0.632. Because one K-value lies above one and the other below, the mixture can split into two phases at this pressure. Solving the Rachford–Rice equation gives β = 38/77 ≈ 0.4935, from which x_A = 0.5 / (1 + β(K_A − 1)) ≈ 0.3889 and x_B ≈ 0.6111, while y_A = K_A x_A ≈ 0.6140 and y_B ≈ 0.3860.

Table 1 shows the compositions and the first three checks:

Both phase sums equal one, and the K-consistency check passes: 0.6140 / 0.3889 ≈ 1.5789 and 0.3860 / 0.6111 ≈ 0.6316. For the component balances, the ledger below uses vapor moles βF ≈ 0.4935 and liquid moles (1 − β)F ≈ 0.5065 on the 1 mol feed basis:

Each component column closes against its 0.5 mol in the feed, and the total column reproduces the phase split. The Rachford–Rice residual check also succeeds: at β ≈ 0.4935 the two terms are 0.2895 / 1.2857 ≈ +0.2251 and −0.1842 / 0.8182 ≈ −0.2251, canceling to zero within rounding. Five independent checks confirm one consistent result.

A phase fraction is not a phase composition

The most frequent misreading of a flash result treats β as if it described the vapor’s makeup. It does not. In a documented Peng–Robinson case at 280 K and 2 MPa — a cold, moderate-pressure condition — the recorded vapor fraction is about 0.65756, while methane’s mole fraction in the vapor is about 0.70979. The two numbers have different denominators: β divides vapor moles by total feed moles, whereas the methane vapor composition divides methane moles in the vapor by all vapor moles. Similar-looking decimals can therefore carry entirely different meanings, and one cannot be derived from the other without the full compositions and balances. This distinction is the central lesson of PetroCalcHub’s guide to phase fraction versus phase composition.

The same discipline applies downstream. Mole fractions are not mass fractions: because phases contain components with different molecular weights, converting between mole and mass fractions requires an explicit molecular-weight step, and mixing the two silently corrupts balances and property estimates. Keep β, the composition vectors x_i and y_i, and the component inventories βF y_i and (1 − β)F x_i as separate quantities, converting or scaling only when a specific question — how much methane ended up in the vapor? — is actually asked.

Common failure modes and remedies

• One composition sums to one, the other does not. The reported x and y likely came from different iterations, or the K-values used in the final pass were stale. Recompute K_i = y_i/x_i from the reported compositions and re-solve once more.
• A trivial solution, x_i = y_i = z_i, or β pinned at 0 or 1. At such points the balance is satisfied for any β, so the two-phase result is spurious. Run a tangent-plane stability analysis to decide whether the feed is genuinely single-phase at the specified T and P.
• Negative compositions or near-zero denominators 1 + β(K_i − 1). The solver has converged to an inadmissible root; admissible Rachford–Rice roots keep all denominators and compositions positive.
• Mole balances close but mass balances do not. The result is sound in moles but was converted to mass without molecular weights, or mole and mass fractions were mixed in one table.
• The wrong specification was solved. Flash variants iterate differently — a bubble-point calculation at fixed pressure iterates on temperature, for instance — and a P–H flash needs an energy balance on top of the component balances. Material balance closing while energy balance does not is a sign the wrong problem was solved.

Practical takeaways

Check a flash result the way an auditor checks a ledger. Verify that both phase compositions sum to one independently; verify that βF y_i + (1 − β)F x_i returns the feed composition component by component; verify that the reported β makes the Rachford–Rice residual vanish with consistent, positive K-values and compositions; and confirm that the specification solved — P–T, bubble point, dew point, or P–H — is the one intended. Keep phase fractions and phase compositions in separate mental columns — β is an amount of vapor, y_i describes what that vapor contains — and this checking habit will catch most solver errors, stale iterations, and basis slips before they reach a flowsheet or a report.