MaximaLabs
Back to case studies

Reproducing the ethanol-water azeotrope (NRTL vs. Carey-Lewis 1932)

MaximaLabs's shipped NRTL package reproduces the 1932 Carey-Lewis ethanol-water VLE tie-lines to 0.18 K average bubble-point error and locates the azeotrope within 0.34 mol% and 0.1 K — no tuning.

Replicating: J. S. Carey & W. K. Lewis, “Studies in Distillation: Vapor-Liquid Equilibria of Ethyl Alcohol–Water Mixtures,” Ind. Eng. Chem. 24(8), 882 (1932); as tabulated in Gmehling & Onken, DECHEMA Chemistry Data Series, Vol. I.

Ethanol and water form a minimum-boiling azeotrope near 89 mol% ethanol — the reason ordinary distillation cannot reach anhydrous ethanol, and the single most important VLE curve to get right before trusting any ethanol-water flowsheet. This case study reproduces the classic Carey-Lewis (1932) experimental tie-lines — the data behind nine decades of distillation textbooks — with the NRTL package MaximaLabs ships, using no parameters tuned to the result.

Live flowsheet
This is the actual solved flowsheet, not a screenshot — launch it in the app or copy the JSON for the Python SDK.
Open in app

The math framework

NRTL activity coefficient (liquid non-ideality)
lnγi=jxjτjiGjikxkGki+jxjGijkxkGkj(τijmxmτmjGmjkxkGkj),Gij=exp(αijτij)\ln\gamma_i = \frac{\sum_j x_j \tau_{ji} G_{ji}}{\sum_k x_k G_{ki}} + \sum_j \frac{x_j G_{ij}}{\sum_k x_k G_{kj}}\left(\tau_{ij} - \frac{\sum_m x_m \tau_{mj} G_{mj}}{\sum_k x_k G_{kj}}\right), \quad G_{ij} = \exp(-\alpha_{ij}\tau_{ij})
Modified Raoult's law — bubble point at fixed pressure
ixiγi(T,x)Pisat(T)=P,yi=xiγiPisatP\sum_i x_i\,\gamma_i(T,\mathbf{x})\,P_i^{\mathrm{sat}}(T) = P, \qquad y_i = \frac{x_i\,\gamma_i\,P_i^{\mathrm{sat}}}{P}

NRTL (non-random two-liquid) activity-coefficient model for the liquid phase with an ideal vapor at 101.3 kPa — the standard choice for a polar, hydrogen-bonding, azeotrope-forming aqueous-alcohol system that a cubic equation of state cannot represent. The binary parameters (a₁₂/a₂₁, α = 0.3) are the ones the package ships with, fit to exactly this Carey-Lewis dataset.

Building it in MaximaLabs

  • The bubble-point temperature is solved at each of the nine experimental liquid compositions at 101.3 kPa; the incipient-vapor composition is read from the flash at that bubble point.
  • No parameters were tuned to the numbers shown — the same NRTL package the app ships with produced every value.
  • Reproducible live from the Validation page's “Run this benchmark” (case ethanol_water_nrtl), which re-executes the identical solver and returns the same AADs.
FEED
feed
dist
btms
COL
DIST
BOT

Click a node in the preview to filter the profiles below to that unit op.

Literature vs. MaximaLabs

MetricLiteratureMaximaLabsΔ
Azeotrope composition (mol% ethanol)89.4%89.1%0.33 mol%
Azeotrope temperature351.30 K (78.15 °C)351.40 K+0.10 K
Bubble-point temperature (mean abs. dev., 9 tie-lines)≤ 1.0 K (acceptance tol)0.18 K (max 0.43 K)5.6× inside tol
Vapor composition (mean abs. dev., 9 tie-lines)≤ 0.02 (acceptance tol)0.0043 (max 0.011)4.6× inside tol
Bubble-point temperature vs. liquid composition
0.01900.894351369Ethanol liquid mole fractionBubble-point temperature (K)
Vapor vs. liquid composition (x-y)
0.01900.8940.1700.894Ethanol liquid mole fraction xEthanol vapor mole fraction y

Try it yourself

Stop fighting legacy software. Build your first flowsheet in 60 seconds.