3D Batch Distillation of Ethanol and Water

A glassware distillation rig with a Vigreux column and Liebig condenser: Antoine and Van Laar equilibrium, plates, reflux, azeotrope and heat balance.

Switch to 2D

Presets

Playback

Parameters

Camera

Layers

Quality

Model check

The science

What is modelled

A batch of ethanol and water is heated in a round flask. The vapour climbs a Vigreux column, is condensed in a Liebig condenser and drips into the receiver. Because ethanol is more volatile, the vapour is richer in ethanol than the liquid, and every plate in the column enriches it again. The page computes this from vapour pressures and liquid non-ideality, not from a lookup table.

Equations

  • Vapour pressure of each pure liquid, Antoine: log10 p = A − B / (T + C) (mmHg, °C), constants from standard tables.
  • Ethanol-water is far from ideal; the Van Laar model gives activity coefficients ln γ1 = A12 [A21x2 / (A12x1 + A21x2)]² (and the mirror formula for water) with A12 = 1.6798 and A21 = 0.9227.
  • Boiling point of a mixture: x γ1 p1(T) + (1 − x) γ2 p2(T) = P, vapour composition y = x γ1 p1 / P.
  • Column at total reflux with N theoretical plates: x1 = y(xflask), xk+1 = y(xk); the distillate is y(xN). The staircase in graph 2 is exactly this recursion; it stalls at the azeotrope (about 89 mol % ethanol in experiments, 91.5 in this Van Laar fit), where y = x and no column can separate further.
  • Batch (Rayleigh) distillation: d(n x) = yd dn, i.e. dx / d ln n = yd − x. For constant relative volatility α it has the closed form ln(W0/W) = [ln(x0/x) + α ln((1 − x)/(1 − x0))] / (α − 1).
  • Heat balance: C dT/dt = η Pel − h (T − 20 °C) until boiling; then the boil-off is V = (η Pel − loss) / ΔHvap and the collection rate D = V / (1 + R) for reflux ratio R.

Checks run in the page

  • The model's pure-ethanol and pure-water boiling points and its azeotrope are compared with literature values (78.3 °C, 100.0 °C, 89.4 mol % at 78.2 °C; the fitted model is about 2 mol % high).
  • An RK4 integration of the Rayleigh equation with constant α is compared with its closed-form solution.
  • The stepped heat balance gives the time to start boiling, compared with the analytic exponential; a whole batch is run to dryness and moles and ethanol are checked for conservation.

Honest simplifications

  • The column is treated at total reflux for its enrichment while the takeoff uses the chosen reflux ratio; a real column at finite reflux separates somewhat less, and plate efficiency, hold-up and column heat losses are ignored.
  • Volumes are added without mixing contraction, heat capacity is a mole-fraction average, the heat loss is one linear coefficient and there is no superheating or foaming.
  • Van Laar constants are a fit at about 1 atm; at other pressures the same constants are reused, so the azeotrope shift is only qualitative.
  • Bubbles, drops and vapour are illustrations; their count follows the computed rates, but not their physical size or speed.

What to try

  • Set 0 plates and 0 reflux: the distillate is only slightly richer than the flask; add plates and reflux and watch the staircase climb towards the azeotrope.
  • Use the preset Reach the azeotrope with 12 plates: the distillate flattens near 91 mol %, however many plates you add.
  • Lower the pressure to 0.5 atm: every boiling point drops and the azeotrope moves (graph 1).
  • Raise the heater power: the boil-off rate rises almost linearly, but a high takeoff with little reflux gives a poorer distillate.
Loading model...
Loading Blender model...
Temperature vs composition (liquid and vapour curves)
Vapour vs liquid composition with the plate staircase
Distillate purity and ethanol recovery vs volume collected
Made with 🟠 Blender · Three.js