Beggs–Brill vs Homogeneous Flow: What the Models Assume

Beggs–Brill and a homogeneous gas-liquid model can predict different pressure changes even when they receive the same pipe and inlet state. A central difference is how they represent the liquid occupying the pipe. The homogeneous model imposes one shared velocity, while Beggs–Brill uses an empirical holdup relationship. Friction closure and treatment of changing properties also differ in the tools compared here.

A useful comparison starts by matching the boundaries rather than running two unrelated defaults. This lesson uses a synthetic 10-m uphill segment with the same diameter, mass rates, inclination and inlet pressure in both models. The inlet gas density is derived from the homogeneous gas-state assumption and supplied explicitly to Beggs–Brill. The comparison explains model behavior; it is not a measured-data accuracy ranking.

At one local state, liquid volumetric flow is QL = mL/rhoL and gas volumetric flow is QG = mG/rhoG. The no-slip liquid fraction is lambdaL = QL/(QL + QG). It describes the liquid share of the combined volumetric flow. In-situ liquid holdup HL describes the fraction of pipe volume occupied by liquid. These two fractions need not agree when the phases move at different mean velocities.

If both phases share one velocity, each phase's flow share equals its occupied cross-sectional share and HL equals lambdaL. That is the homogeneous no-slip assumption. Beggs–Brill instead estimates holdup using its reference regime map and inclination treatment. Its reported map label is a correlation category; it is not direct observation of a slug, bubble or annular pattern in a real inclined pipe.

The shared geometry is a 0.05-m inside diameter and 10 m of along-flow length, inclined 30 degrees uphill from horizontal. Gas and liquid mass rates are 0.05 and 0.55 kg/s. Liquid density is 850 kg/m³. Inlet absolute pressure is 2 MPa. The homogeneous model uses constant temperature 320 K, molecular mass 20 kg/kmol and Z = 0.9.

Those gas inputs imply inlet density rhoG = pM/(ZRT) = 16.704493756 kg/m³ with R = 8,314.462618 J/(kmol·K) and pressure in Pa. Supply that same density to Beggs–Brill rather than retaining its unrelated default gas density of 20 kg/m³. Both models then start from lambdaL = 0.177750455, approximately 17.775%. Their common inlet mixture velocity is about 1.853972 m/s.

At the baseline angle, the Beggs–Brill short-segment calculation predicts a pressure loss of 18.051362 kPa. Its ledger contains 1.537382 kPa of friction, 16.505393 kPa of gravitational loss and 0.008587 kPa of acceleration contribution. Its liquid holdup is about 0.383911648, above the common inlet no-slip liquid fraction of 0.177750455.

The homogeneous profile gives 9.203878 kPa total loss, with 1.135216 kPa integrated friction, 8.066508 kPa integrated gravity and 0.002154 kPa acceleration. It updates gas density along the pressure profile at fixed T and Z, while this Beggs–Brill segment freezes its local properties. The difference therefore includes holdup, friction treatment and property evolution; it cannot be attributed solely to one slip correction.

The homogeneous workflow accepts a constant Darcy friction factor. Here it is deliberately supplied as 0.02 rather than inferred from an unspecified mixture viscosity. The Beggs–Brill implementation solves a smooth-pipe friction relation using no-slip properties and applies its two-phase multiplier; the baseline resulting two-phase Darcy factor is about 0.0271367. These are different closures even at the same inlet state.

The explorer lets you change the homogeneous supplied factor while leaving the Beggs–Brill closure alone. That isolates one declared assumption instead of silently making both curves agree. Keep Darcy and Fanning conventions distinct: a Darcy factor is four times the corresponding Fanning factor. Matching a numerical factor without matching its convention would create a comparison error before either flow model was evaluated.

The baseline no-slip Reynolds number is about 80,380, within this Beggs–Brill implementation's accepted turbulent range of 4,000 to 10,000,000. Its acceleration parameter is about 0.000476 and its segment pressure change is less than 1% of the 2-MPa inlet pressure. The code also checks holdup and friction conditioning. These checks make the selected numerical case acceptable to the software; they do not certify empirical accuracy.

The homogeneous model integrates friction and gravity and checks the independently evaluated acceleration contribution against its total momentum balance. It is limited to supported uphill or horizontal states and avoids its critical-flow boundary. The explorer confines inclination to 0–60 degrees and the supplied Darcy factor to 0.01–0.04; every selectable combination is checked in the lesson tests. A longer interval would need a fresh assessment of the frozen-property Beggs–Brill segment.

A no-slip model is useful for understanding conservation, property changes and the consequences of a supplied friction assumption. Beggs–Brill introduces an empirical holdup and regime-based treatment for a supported local segment. The implementation references linked here document the particular correlation choices; they are not a claim that the original experimental paper or a field dataset validates this synthetic case.

Use representative pressure and holdup measurements when judging predictive performance for a real application. Numerical closure, agreement between two software implementations and agreement with measurements are different forms of evidence. Neither workflow here simulates severe slugging, a complete thermal/compositional system or an automatically calibrated well model. If phase slip is central to the question, the linked vertical drift-flux workflow provides another explicitly scoped closure to study.